Set CI buildtypes to highest opt level possible and set default buildtype to debugoptimized

[kazarmy]

Your checklist for this pull request

• I've read the guidelines for contributing to this repository
• I made sure to follow the project's coding style
• I've added tests that prove my fix is effective or that my feature works (if possible)
• I've updated the documentation and the rizin book with the relevant information (if needed)

Detailed description

This pr:

1. Sets the buildtype for CI builds to release (-O3) except the asan and Codecov builds. This is done because compiler optimizers increase performance but can also change code behavior (especially if the code invokes UB) so this needs to be tested.
   1.1 The asan buildtype is set to debugoptimized (-O2 -g) for hopefully sane stack traces.
   1.2 The Codecov buildtype is set to debug because debugoptimized (as seen in Set CI buildtypes to highest opt level possible and set default buildtype to debugoptimized #260 (comment)) produces wonky Codecov reports.

2. Set the default buildtype to debugoptimized due to reasons given in Set default meson buildtype to debugoptimized? #256 i.e. users might leave performance on the table otherwise.

Test plan

All meson builds are green, including Windows and asan builds.

Closing issues

Closes #256.

[codecov-io]

Codecov Report

Merging #260 (06fba52) into dev (b37d6fb) will decrease coverage by 0.01%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##              dev     #260      +/-   ##
==========================================
- Coverage   42.80%   42.79%   -0.02%     
==========================================
  Files         871      871              
  Lines      316886   316550     -336     
==========================================
- Hits       135653   135457     -196     
+ Misses     181233   181093     -140     

Impacted Files	Coverage Δ
librz/asm/p/asm_dcpu16.c	0.00% <ø> (ø)
librz/asm/arch/hexagon/hexagon_disas.c	0.11% <100.00%> (ø)
librz/bin/format/wasm/wasm.c	40.28% <100.00%> (ø)
shlr/gdb/src/gdbclient/responses.c	15.03% <0.00%> (-18.43%)	⬇️
librz/asm/p/asm_pyc.c	0.00% <0.00%> (-14.29%)	⬇️
librz/core/cplugin.c	65.38% <0.00%> (-13.57%)	⬇️
librz/bin/format/pyc/pyc_magic.c	0.00% <0.00%> (-12.83%)	⬇️
librz/bin/p/bin_pyc.c	0.00% <0.00%> (-7.32%)	⬇️
librz/reg/reg.c	77.77% <0.00%> (-5.26%)	⬇️
librz/util/unum.c	71.48% <0.00%> (-5.08%)	⬇️
... and 778 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update b37d6fb...2a5d47a. Read the comment docs.

[ret2libc]

why this?

[kazarmy]

because of (https://github.com/rizinorg/rizin/runs/1608321103#step:12:2082):

I'm assuming gcc ran out of memory (based on e.g. https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87316#c3).

[ret2libc]

did you remove -Werror just for test?

[kazarmy]

yes, it should be reenabled at the end

[ret2libc]

I'm ok overall with the idea of moving to debugoptmized.

[kazarmy]

Not sure whether Codecov output is reliable under optimization though.

[kazarmy]

Not sure whether Codecov output is reliable under optimization though.

Unfortunately, it appears that lines can disappear alarmingly from the Codecov report under -O2 optimization. Compare e.g. a report for librz/parse/p/parse_ppc_pseudo.c on -O0:

and on -O2:

especially the line on which the report restarts.

[ret2libc]

@kazarmy we can still default to debugoptimized but use debug in the codecov CI job.

[kazarmy]

I did a separate build for Codecov 6a1af99 since didn't want to lose the running of tests on Linux with debugoptimized.