Add support for subtype matching in patterns.

[rofinn]

We sometimes use types in our patterns, so this would allow us to use parent types or Unions to describe a collection of paths to match against.

[codecov]

Codecov Report

Merging #47 (b386dd7) into main (d39b647) will not change coverage.
The diff coverage is 100.00%.

@@           Coverage Diff           @@
##             main      #47   +/-   ##
=======================================
  Coverage   98.88%   98.88%           
=======================================
  Files           7        7           
  Lines         270      270           
=======================================
  Hits          267      267           
  Misses          3        3           

Impacted Files	Coverage Δ
src/patterns.jl	100.00% <100.00%> (ø)

---

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 d39b647...b386dd7. Read the comment docs.

[glennmoy]

Is there any chance of there being an ambiguity?

For instance, if some tuple element contains two types that are both subtypes of T, is there some Pattern that could be given that would not be able to resolve which one to choose? Or does it always pick the first?

I tried pseudo-coding an example but wasn't able to come up with one, so perhaps it's not possible, or maybe it's friday evening brain haze.

[glennmoy]

FYI you might want to change the repo settings to only allow merge on approval.

[rofinn]

Same as with the wildcards it'll just pick the first match, which is why we have an OrderedSet to have consistent behaviour. This will likely make the solving #36 a bit harder, but probably worth it.