tests/fixedbugs: regression test for issue 100

[mmcloughlin]

Adds a regression test based on klauspost/compress#186.

Includes tweaks to other regression tests to ensure they are run correctly in CI.

Updates #100 #65

[codecov-io]

Codecov Report

Merging #129 into master will increase coverage by 0.41%.
The diff coverage is 94.87%.

@@            Coverage Diff             @@
##           master     #129      +/-   ##
==========================================
+ Coverage   77.63%   78.04%   +0.41%     
==========================================
  Files          55       55              
  Lines       33376    33391      +15     
==========================================
+ Hits        25910    26059     +149     
+ Misses       7353     7185     -168     
- Partials      113      147      +34

Flag	Coverage Δ
#integration	8.38% <100%> (+1.97%)	⬆️
#unittests	75.62% <69.23%> (-0.01%)	⬇️

Impacted Files	Coverage Δ
internal/gen/ctors.go	0% <0%> (ø)	⬆️
printer/goasm.go	95.45% <100%> (+0.58%)	⬆️
pass/cfg.go	100% <100%> (ø)	⬆️
x86/zctors.go	95.97% <100%> (ø)	⬆️
ir/ir.go	90.62% <100%> (+0.14%)	⬆️
internal/prnt/printer.go	66.66% <0%> (-11.12%)	⬇️
build/global.go	82.6% <0%> (+2.17%)	⬆️
build/zinstructions.go	5.61% <0%> (+2.23%)	⬆️
build/context.go	63% <0%> (+5%)	⬆️

---

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 5a144d9...c641099. Read the comment docs.

[mmcloughlin]

The lint job failed because of differences with asmfmt. Specifically asmfmt has the following features that we don't implement:

• Instruction printing is reset after a JMP instruction
• Parameterless instructions do not contribute to the "max instruction length" of a block https://github.com/klauspost/asmfmt/blob/3da6afce54d96eb4f7ec7ffaa196f53152e9d456/asmfmt.go#L558-L559

Related #8