Fix extended trait change issues

[corranwebster]

This fixes #537 and fixes #538.

It also removes expected failures of tests.

This may break existing code that accidentally uses buggy behaviour.

[codecov-io]

Codecov Report

Merging #621 into master will increase coverage by 0.33%.
The diff coverage is 60%.

@@            Coverage Diff             @@
##           master     #621      +/-   ##
==========================================
+ Coverage   69.43%   69.76%   +0.33%     
==========================================
  Files          40       40              
  Lines        6282     6285       +3     
  Branches     1295     1296       +1     
==========================================
+ Hits         4362     4385      +23     
+ Misses       1521     1501      -20     
  Partials      399      399

Impacted Files	Coverage Δ
traits/traits_listener.py	80.04% <60%> (+0.95%)	⬆️
traits/trait_types.py	66.33% <0%> (-0.17%)	⬇️
traits/has_traits.py	70.3% <0%> (+0.16%)	⬆️
traits/traits.py	61.25% <0%> (+1.29%)	⬆️
traits/etsconfig/etsconfig.py	63.58% <0%> (+6.17%)	⬆️

---

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 761d76d...74c2179. Read the comment docs.

[mdickinson]

As a data point, this is causing a significant number of test failures on a largish Traits-using project; some repetitions, but it looks like 10-15 different instances of failure.

[mdickinson]

Given the extent of the breakage seen in testing, I don't think we can get this into 6.0.0. We need to find some way to provide a non-breaking upgrade-path for Traits-using applications.

[mdickinson]

@corranwebster Thoughts on what to do with this PR? We can leave it open and use it as a starting-point for more far-reaching updates. Or should it be closed?

[corranwebster]

I think it needs to be closed sooner or later: I think we've demonstrated that too much downstream code will be broken by this change.

Certainly we don't want an accidental merge.

[mdickinson]

Closing here; I've left a couple of comments on #537 and #538 pointing to this PR and explaining why it wasn't used as the fix.