Over the course of your dev log, it seems you asked or at least opened up a series of topics for discussion. In no particular order, here is some of feedback for some of those topics.
As with any advice I give, feel free to use it, if it helps; discard it, if it does not. These are the thoughts of one individual trying to learn as much as possible and improve.
Class systems:
As far as class swapping was concerned, there were two instances I found it important.
First, allowing the player to select their class, so that they have a character in game that fits their play style. This was something I initially ignored but later added in as extra content to one of the DLCs where it was lacking.
Second, showing another side to a character or their progression. There is a hidden side quest I added into a game that allows one of the characters to radically change classes. It foreshadows a role they will settle into later in life. It can also be used to show another side to a character, I made two classes [one of which I haven't released, yet] and revealing that a character can use both would show the nature of their power. I likely won't use the unreleased class in this manner as it might be contradictory to what that character did previously and I have follow-up quests already written where the reveal was unnecessary [it wouldn't be impossible, only more difficult but possibly intriguing if done right].
On writing:
This is in reply to both the intro cutscene length and total length of a game.
As far as writing an intro goes, there are some devs that rely entirely on their writing to keep a player invested. Others will push the player in as quickly as possible. What matters is cutting excess off the dialogue. More often than anything, the excess is what makes dialogue self indulgent. Despite my efforts, there are parts that need trimming in my own works. How do I know when it's excessive? When it doesn't setup future events, explain/show relevant motivations, or move the story forward. A few extra lines here or there shouldn't hurt but I have a mantra for this: “Don't waste the player's time”.
I know this might be brutal, but you have to kill your darlings. I've built up history, lore, cultural, and economic systems but none of it matters to most players. I didn't completely remove it from the game but the most detailed parts that shape the world itself are left in the margins.
I try to keep the narrative relatively tight, closer to a movie script. I agree with Ross Scott on this one; in his Fahrenheit review, he noted that if not executed properly, the choices will seem meaningless. But there are other games that have given a handful of choices and pulled it off masterfully [Eternal Darkness developed by Silicon Knights comes to mind].
As far as total length, that will depend on the content of the story. In my case, I turned one quest into a side quest and, in another instance, I removed one of the missions from the main campaign and made it into an mini epilogue arc. I've written character interactions that can entirely be skipped, even though it affects future events. In these cases, the content is there if the player wants it but it can also be skipped as to not waste the player's time.
Some may even argue that you don't even need much of any story or character progression. “Are you a bad enough dude to rescue the president?” But that is more common in action games.
That said, there's nothing inherently wrong with lengthy cutscenes.
Regarding Combat Systems:
I still think turn based systems are viable.
Battletech [2018] used a CTB system, if I recall correctly, and that seemed to sell well enough. Players will play turn based games, they need to be sold on it [I need to work on this]. As to what works for your game, that will depend the mechanics involved.
For myself, I went with an STB system as it worked best with design choices I made for the classes. Two of the [released] classes have mechanics where their skills have an override in action order. This would be the Guardian with their taunts [and this applies to the Vanguard and Defender subclasses as well] and the Bard skills Vulnerability and Syren's Song [which can be countered by another Bard using Opposing Tune]. When testing the game with ATB and CTB systems, these skills became broken.
Additionally, the standard ATB and CTB systems in MZ use agility as the determining stat for turn speed. The problem I had with this is that using the stat distributions I had, this would encourage players to seek out or roll characters with a high agility stat over others. This quickly started slide into a downward spiral of calculating what would happen as a result. In addition to breaking class skills here are a few more examples of possible corrections, I could make it so that the classes had about the same agility [removing statistical distinction between classes], recovery items could be reduced, in some way, to force players to bring a support class along. Eventually, I realized that one of the end results would be degrade the classes to damage dealers spamming health potions or having the token healer in tow. There's nothing wrong with that but it could be still done with an STB system and it wouldn't compromise the gameplay experience for other players choosing other play styles.
I'd be open to using these systems in the future but I'd have to make a different game to fit these mechanics.
One thing I kept in mind was to give the player considerations and trade offs for each action taken. Do I use a standard attack to gain to fill up my skill bar faster? Do I use a mana based skill and have a better chance of it working but have my skill point meter possibly rise slower? I was streaming as part of a promotional effort this last week. During the streams, I found myself having to take a moment and think about my actions and even going back to change some actions mid turn to optimize my strategy. And these are mechanics that I've beaten to death during play testing and recreational play throughs.
On Performance:
As for performance, I made it so that my games could be comfortably played on even a budget netbook. I went so far as to add a free DLC to make sure it would run at an average of 30 FPS on said netbook [mostly an MZ issue with the fancy combat animations]. Most PCs seem to run it fine at 60 FPS without the optimization patch.