Provide an inline asm fallback for the ARMv8 intrinsics

[ccawley2011]

The configure checks have also been modified to better match the ARMv6 code.

[codecov]

Codecov Report

Attention: Patch coverage is 66.66667% with 1 lines in your changes are missing coverage. Please review.

Project coverage is 83.04%. Comparing base (93b870f) to head (cc24587).
Report is 9 commits behind head on develop.

Files	Patch %	Lines
test/test_crc32.cc	0.00%	0 Missing and 1 partial ⚠️

Additional details and impacted files

@@           Coverage Diff            @@
##           develop    #1697   +/-   ##
========================================
  Coverage    83.03%   83.04%           
========================================
  Files          134      134           
  Lines        10336    10336           
  Branches      2813     2813           
========================================
+ Hits          8583     8584    +1     
- Misses        1054     1057    +3     
+ Partials       699      695    -4     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[mtl1979]

Renaming ACLE to just ARMV8 is misleading... ARMv8 is a lot more than just ACLE CRC instructions.

[nmoinvaz]

I'm not familar enough with ARM, why rename ACLE to ARMv8?

[ccawley2011]

I'm not familar enough with ARM, why rename ACLE to ARMv8?

ACLE is the name of the header for non-NEON intrinsics rather than a specific CPU extension. It's used for both ARMv8 CRC32 instructions and ARMv6 SIMD instructions, however the build system for zlib-ng only uses ARM_ACLE to refer to the ARMv8 code. The ARMv6 code uses a separate option.

[mtl1979]

We can't just rename command-line options for configure or CMake build stage, as those have to be backwards compatible. Otherwise we break build for every project that use zlib-ng. Adding support for inline asm fallback should be separated as own pull request, as it is less controversial, even though we have no precedence for supporting mismatched toolchains, which is clearly the case here as GNU assembler would be newer than GNU C compiler (GCC).

[Dead2]

Renaming should be split into a separate PR from the added features/fixes.

[Un1q32]

The improved acle/armv8 check fixes a false positive when building for 32 bit arm iOS, I ripped out the improved check and put it in https://github.com/Un1q32/zlib-ng/tree/acle, I would make a PR with just that but I'll wait to see if the author of this PR finishes the main PR

[Un1q32]

No activity here for a while, gonna make a PR for the bit I ripped out