Some of your "wants" seem to be inconsistent, mutually exclusive and/or contradictory.
For example, you talk about Invulnerable to coercion from world powers but then ask for "smartphone"....
You may know that back a year or so ago Tim Cook came out of the woodworks and came out of the closest and disclosed that Apple has never built any backdoor for NSA and etc.... then they said they were going to starting OS 8 do FDE on the entire phone. What they didn't tell the public is that all phones are belong to the NSA/GCHQ in the first place a la Gemalto type SIM card exploits. Also, what the masses don't know is that them using a four digit code to protect their device is about as good as not having one at all. No cellphone or smartphone is secure nor can be reasonably securable. Plus even the form factor doesn't lead itself to security. Unlocking phone with swipes? Come on. And you will never be able to enter a 64 character FDE passcode or use Yubikey static passcode to unlock a cell phones FDE like the way you could do so with TC7.1a on the desktop PC. It is not practical at all and the form factor doesn't lend itself to that. Not to mention these FDE implementations are from Apple and Google, and can you really trust these known PRISM partners? You did after all talk about " coercion from world powers:" did you not? You can't have it both ways.
Prediction; the day Google Android, Apple or Samsung will support native FDE or Encryption with "Fake containers" either for plausible denialability or whatever other reasons : NEVER.
"I am not going to get a Ph.D trying to figure out how to use it." - you don't have to have a PhD to figure out how to use TrueCrypt. It is fairly easy and there are plenty of documentation available and it is fully open source unlike bitlocker and it has already been audited twice (for whatever that is worth). So read all of that. So you can get intimately familiar with TC without any degree in anything. Conversely, having a PhD alone doesn't mean you know anything about encryption or security.
So I approach this whole thing differently. From the exact opposite angle that you have. Instead of asking most suitable alternative to TrueCrypt, I ask myself whenever a situation arises: will it pass the TrueCrypt 7.1a litmus test.
If I can only buy a computer that is locked down to UEFI and I can't do emulated legacy BIOS and I can't install TC7.1a on it, then it is a "nogo" and I don't buy that computer. (ie staring this summer, Windows 10 debuts and OEM have the option of completely locking down SecureBoot/UEFI so that you CAN'T boot to anything else nor install any other OS, TAILS, TOR, Linux, Windows 7, etc) If that is the case, then I stay away from that and I don't like the trend of the desktop becoming locked down like a cell phone. There are still plenty of options and I can build my own PC and there will always be motherboards around that are unlocked.
If a particular OS only supports GPT or if a particular harddrive, drive, or partition and/or volume is GPT then that is no go for me. Either I wipe it clean and reformat it as MFT, or I go with some other OS, firmware, drive, combo, etc that works for MBR/MFT and that I can effectively install TC7.1a FDE on.
The issue that VeraCrypt tries to solve "TrueCrypt uses 1000 iterations of the PBKDF2-RIPEMD160 algorithm for system partitions, VeraCrypt uses 327,661 iterations" is a nonstarter for me. If you are going to be using TC/TCFDE then you should be using a long and secure enough password that it wouldn't matter if it had 3million, 1 thousand or just 1 iteration....
Since my FDE passcode is 64 character long, I don't worry about it.
The second TC audit exposes a potential vulnerability of "AES Implementation susceptible to cache timing attacks" and recommends stop using TC's software based AES and trusting Intel's AES-NI instead. On for example my gaming computer I use Hardware accelerator because of speeds. Not that I trust Intel. But on any important computer you would want to at the very least airgap it and then use TC FDE with Triple Cascading ciphers. And I would not trust Intel AES-NI not to have some backdoors. So again, the issue is a nonstarter. The audit recommendation for most people to steer away from using Triple cascading ciphers makes me thing of it as very fishy.
Is TC perfect? No. Some might think it is dated and aging. But as far as I'm concerned, it is better to approach it from the perspective of can this computer or computer system pass the TrueCrypt 7.1a FDE litmus test ? If not, then ask yourself why? Probably good reason to stay clear or at least reconsider your other options.
Perfect security at a push button does not exists.