More runtime formatting

[IntegratedQuantum]

bufPrint and allocPrint at runtime

Should reduce compile time and binary size, and reduces catch unreachable

follow-up on #2792

Perfomance:

.	Before	After	Binary size
Debug + llvm	13.75 s	13.3 s	- 0.3 MB
Debug + x86_64	4.85 s	4.75 s	+ 12.4 MB (the whole binary is over 600 MB, probably for incremental)
ReleaseSafe	82.8 s	81.2 s	- 0.4 MB
ReleaseFast	90.0 s	85.5 s	- 0.8 MB

Remaining work:

• Make performance comparisons
• Check if it catches mistakes at compile-time when running tests
• There is a problem with vectors and precision

[IntegratedQuantum]

I have decided to discard this for now, since the x86 backend is now usable, it doesn't provide a meaningful compile-time performance improvement, and there are many edge-cases that aren't handled correctly (e.g. printing characters)