Fix regression on anonymous namespaces.

[SoftwareApe]

I remembered this morning that there was a special case on anonymous namespaces in C++, and indeed I had accidentally removed that. Since there was no unit test for it I added a unit test for anonymous namespaces on tidy_generic so this regression shouldn't happen again.

While at it I saw that there was a different kind of special case stack entry in C++ with lambdas written as {lambda(int, int)#2}, which again got truncated. So I also made the bracket detection in tidy_generic more generic, of course with tests 😄.

[codecov]

Codecov Report

Merging #175 into master will increase coverage by 0.04%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master     #175      +/-   ##
==========================================
+ Coverage   90.19%   90.23%   +0.04%     
==========================================
  Files          16       16              
  Lines        2224     2223       -1     
==========================================
  Hits         2006     2006              
+ Misses        218      217       -1     

Impacted Files	Coverage Δ
src/collapse/perf.rs	97.36% <100.00%> (-0.01%)	⬇️
src/collapse/common.rs	62.87% <0.00%> (+0.33%)	⬆️

---

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 abd628f...5084bc1. Read the comment docs.

[jonhoo]

Oh, nice catch!
This does assume that brackets are balanced, but that seems like an okay assumption to make.