Identify invalidating blockers of precompilation

[timholy]

This allows one to identify the particular invalidations that block
precompilation of particular MethodInstances.

I am not 100% sure this is right. In particular, it might be too conservative; maybe one should SnoopCompile.getrroot(hits[i]) to see the real impact? But I checked whether the root CodeInstances are there in the precompile cache, and they seem to be. So my best guess is that this is painting an accurate picture as-is. I might let this sit on master for a few days before making a release.

CC @ChrisRackauckas. It seems to suggest that invalidation is not your first-order problem with SciML/ModelingToolkit.jl#1215, as you seemed to expect. But I wanted to get this tool out there in some form.

[codecov]

Codecov Report

Merging #262 (f991a4e) into master (8b6bc60) will increase coverage by 1.41%.
The diff coverage is 98.27%.

@@            Coverage Diff             @@
##           master     #262      +/-   ##
==========================================
+ Coverage   85.16%   86.58%   +1.41%     
==========================================
  Files          15       16       +1     
  Lines        1807     1878      +71     
==========================================
+ Hits         1539     1626      +87     
+ Misses        268      252      -16     

Impacted Files	Coverage Δ
src/SnoopCompile.jl	100.00% <ø> (ø)
src/invalidations.jl	90.28% <96.55%> (+1.89%)	⬆️
src/invalidation_and_inference.jl	100.00% <100.00%> (ø)
src/parcel_snoopi_deep.jl	89.13% <100.00%> (+1.31%)	⬆️

---

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 8b6bc60...f991a4e. Read the comment docs.

[timholy]

I should say that for this to work well you need to be running at least Julia 1.8.0-DEV.368 (you need JuliaLang/julia#41913)