Improve checking for GCC >= 8 and Clang for passing -Wno-class-memaccess

[mstorsjo]

Don't pass -Wno-class-memaccess to Clang, which doesn't support that option name. (When inspecting $(CXX) with a pattern, "clang++" also matches the pattern "%g++".)

Simplify the pattern checking; use $(filter %g++,$(CXX)) instead of $(patsubst %g++,,$(CXX)) making it less obscure. (This requires changing the ifeq ($(),) into ifneq.)

Check for %clang++ in addition to %g++.

Also use a similar pattern for checking for clang++ instead of plain exact matches for "clang++", for passing the option -Wc++11-compat-reserved-user-defined-literal - this allows applying the option when CXX contains a path, or if "clang++" is prefixed by a cross triple.

[GuangweiWang]

LGTM

[mstorsjo]

All the failures in CI are in ThreadDecodeFile/ThreadDecoderOutputTest.CompareOutput/*, which fails occasionally independent of these changes.

I'm considering if we should entirely ignore the ThreadDecodFile tests in the CI setup, since they're not passing deterministically. That would lose test coverage of them (instead of being mostly-working but failing occasionally, they could break entirely), but currently they're causing a lot of noise.

(Also, why can't the threaded decoding be fixed so that it deterministically always produces the correct reference output, instead of randomly producing different results?)

[huili2]

@xiaotiansf now this series of UT greatly breaks our CI. Do you think it's easy to fix or we'd disable them first?

[xiaotiansf]

I think it is safer to disable it in UT (I noticed there is two test cases 29 and 30 which generate different hash) for now until I find time to look at and fix it.

[mstorsjo]

Now after rebasing past 8535525, CI is passing cleanly.