Right, but that was seriously starting to change by the time the Dreamcast rolled around. DC was such an arcade-centric system (understandably, since the NAOMI board was so hugely successful in arcades), but this is also when the market started to turn away from this sort of stuff.
DC had a strong arcade heritage of course, but let's not act like those were the ONLY type of games on the system. Hell, its most prolific games, the majority anyway, were console-orientated releases. Sonic Adventure, Jet Set Radio, Skies of Arcadia, D2, Shenmue, Grandia II etc...these were games that got the most coverage and most attention for the platform, and were all home-exclusive titles big on content.
Games like Soul Calibur and DOA2 were also very popular at the time, of course, but those were not the majority in terms of individual games which garnered the most attention of the platform. Even so, those games were also packed with a ton of extra home-focused content. As for DC/NAOMI synergy, IMO that was a great strategy by Sega and something they should've done for the Saturn. Maybe they should've made the STV home for games like VF2 and Indy 500 to get near-perfect visual parity with Saturn home ports, and could've spent the time it took getting Model 2 conversions to run on Saturn, on bonus content for the Saturn ports instead if using STV. Then a year or so later do an upgraded port for the Model 2 board with improved visuals.
Model 2 should've been reserved for arcade-exclusive games and visually improved ports of STV/Saturn games coming a year or so after the initial releases, IMHO.
Ehhhhh... Capcom really partnered big with Sega to port almost all of their arcade games at the time to Dreamcast, and while I won't say those ports were barren, I will say a lot of them were major flops. The fact is, it doesn't mattern how many extra modes you put in a game like Mars Matrix or Project Justice (and they did put some lovely extras in those), not many people wanted those games.
Okay but there's also the reality of what marketing budgets were partitioned to those games, and the fact that some like Project Justice never got American ports anyway. Also the quality of extra content would obviously vary from game to game; not every Capcom home port got the level of extras SFA 3 did, for example.
So again, you can't really say that arcade-style games as a whole declined off a cliff from end of 5th-gen heading into 6th-gen, because there were still a lot of games with arcade-style design sensibilities (if the person's concept of those go beyond "lives systems and quarter-munchers", the latter of which was overblown and often the result of the operators configuring the machines' credits system through DIP switches) that did very well in 6th-gen. VF4 and VF4 EVO, MSR, Tekken 4, Tekken 5, the Burnout series, NFS Hot Pursuit 2, Twisted Metal Black, the MK games (the fighting ones, anyway), Maximo, the Beatmania and Dance Dance Revolution titles, Guitar Hero, even IP like Devil May Cry and Viewtiful Joe.
These were all either arcade games with home ports, or games with design sensibilities born out of arcade game lineages (and with some tuning, could've worked very well as arcade releases....in fact some of them DID get arcade releases just further proving the point) that did very well for themselves and their publishers that generation. So the notion arcade-style games "died out" just was never once true. Certain genres tied strongly to arcades DID drop off, mainly shmups and light gun rail shooters, but that wasn't the entirety of arcade games in terms of genres at home to that space. One thing all of the examples I listed though, is that they had a lot of content structured for the home market (including in some cases stronger emphasis on story), and had releases on healthy console platforms.
Which is probably the bigger proof as to why those Capcom games on DC you mention didn't do super-great: Sega's brand name at the time just wasn't very good anymore due to their mistakes with 32X and Saturn (and Mega CD, tho less so in that one's case). They lacked the brand power and marketing muscle to push Dreamcast as big as it deserved to, and while they had a lot of great games in that time period, they never really had the "one" game that was so ahead of what Sony and Nintendo were offering that it shifted all eyes onto Sega. Shenmue was meant to be that game but, it didn't resonate the way they probably hoped in the West, and I don't remember it doing too much for Dreamcast in Japan either though by then it was probably too late due to PS2's release.
Capcom porting CPS2 and CPS3 arcade games to Dreamcast wasn't going to give the system the public profile it needed to bolster the install base, regardless of how good the ports were or what extra content came about. I guess we can also apply this to my other examples, in that the games I mentioned benefited a lot from being on a platform like PS2 and Sony's strong goodwill among the industry at that time period, combined with their 1P output and marketing prowess/muscle giving the install base to let a lot of those PS2 arcade ports and arcade-style games to proliferate and find their audiences, something Sega was never really in a position to do with Dreamcast unless they had a bigger breakout of install base uptick in the late '98 - '00 period, particularly before PS2's American launch.
TBH, the rushed, shitty ports of VF and Daytona had more to do with the bad hardware decisions and rushed launch than anything else. Like I said Sega made a lot of mistakes around that time, beyond simply their choice to go with new IP.
I wouldn't say "bad hardware decisions" as in the hardware design itself; while it was cumbersome, parallel processing wasn't new to Sega's teams by that point, and it's not like the Saturn had critical unfinished hardware flaws (something the Jaguar suffered from) or didn't providing any documentation (they gave full documentation of all hardware from Day 1).
It's just that Sega didn't build a good enough SDK environment, particularly one that leveraged C language, and they did miss some features in hardware despite having the raw power to do them in software (MPEG decoding for example). Truth is they already had a vastly superior port of VF1 for Saturn even by the time of the surprise May launch, because VF Remix was shown behind closed doors at that E3. They simply decided to push out the buggy VF1 instead, saving VF Remix for the September Saturnday launch I guess (Pandemonium Games has a great VF Remix doc on his channel talking about this, worth a watch!).
The worst part of this was SoJ knee capped EC on the Saturn because they didn't want it to affect VF sales. I always point out this was an idiotic move because given when EC CD was released there was no way Saturn EC was going to come out less than a 6 months after VF2. (It couldn't have affected VF2 sales at all given at best it'd come out mid 96 and the vast majority of VF2 sales would have already happened.)
Yeah; I think it was a matter of pride in those days but with certain Japanese publishers in general, teams weren't keen on sharing tech among each other. Sega were one of the worst at this, resulting in the Sonic Extreme cancellation among other things. SoJ seemed really petty over MegaDrive/Genesis and they wanted their "home-grown" IP to be pushed in those territories instead.