MDBF-1195: Define C/ODBC ASAN/UBSAN builders

[RazvanLiviuVarzaru]

As per:

• https://buildbot.mariadb.net/buildbot/builders/codbc-linux-amd64-asan
• https://buildbot.mariadb.net/buildbot/builders/codbc-linux-amd64-ubsan

[grooverdan]

I know its migration but there's a chance to improve the situation:

• ASAN and UBSAN can be instrumented together in the same builder like the server
• gcc is a older and more limited ASAN/UBSAN, tested with the msan-clang-22 container and on the ODBC side anyway - there's ODBC-489 and CONC-813.
• the compiler itself highlight a number of warnings not visible by gcc

The code fixes for these failures aren't extreme, so we could make the block MDBF-1195.

I haven't yet run the pure C/C test on ASAN/UBSAN clang-22.

[RazvanLiviuVarzaru]

Right! I knew you are the go to person on asan/ubsan, thanks for opening the discussion on improvements!
Let's see what Lawrin has to say about.

[grooverdan]

Did test C/C directly - found additionally - CONC-814 - not relevant for this PR however.

note env for the CI env:

export UBSAN_OPTIONS=print_stacktrace=1:report_error_type=1:halt_on_error=1
export ASAN_OPTIONS=detect_stack_use_after_return=1:detect_leaks=1:abort_on_error=1:atexit=0:detect_invalid_pointer_pairs=3:dump_instruction_bytes=1:allocator_may_return_null=1

[grooverdan]

❤️

[RazvanLiviuVarzaru]

As the name suggests, 9c5086e, is just a draft, suppressions won't be hard-coded in Buildbot, should be in the upstream. The list is showing what is required to make the builder green, doesn't mean it is correct to suppress all of them.