Fix axes and size of BlockIndices

[mtfishman]

The current definition of size(iter::BlockIndices) was assuming the indices were ranges (resulting from #483 where BlockIndexRange was generalized to BlockIndices but the definition if size wasn't updated accordingly), so for (non-contiguous) non-ranges it was giving the wrong result, this PR fixes that. Additionally, previously only size was defined which meant that BlockIndices where the indices had non-trivial axes (like blocked axes) lost information, which is now fixed and tested.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 94.53%. Comparing base (7ebbea7) to head (0da26a3).
⚠️ Report is 2 commits behind head on master.

Additional details and impacted files

@@           Coverage Diff           @@
##           master     #490   +/-   ##
=======================================
  Coverage   94.52%   94.53%           
=======================================
  Files          19       19           
  Lines        1789     1810   +21     
=======================================
+ Hits         1691     1711   +20     
- Misses         98       99    +1     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[mtfishman]

@dlfivefifty @jishnub I'll merge this tomorrow unless I hear back from you earlier, I think the definition is uncontroversial and it would be a nasty silent bug to hit.

[dlfivefifty]

I merged it.

I take the view that if (1) nothing downstream is broken and (2) you are confident the changes are good/correct; then please feel free to just merge and tag without review. We can always revert a change if it turns out to be a problem (which often involves adding another downstream test where the problem was detected).