Performance: Memoize multi selection selectors

[aduth]

This pull request is a first pass at render performance for posts containing many blocks. I had noted that simple changes to editor state were causing the entire list of blocks to rerender, which could degrade performance particularly when many blocks exist. With the default behavior of react-redux's connect applying "pure" behavior, it was unexpected that other components should be rerendering aside from the one being hovered.

• Before (2mb GIF)
• After (2mb GIF)

Implementation notes:

In this case, the VisualBlockList component would rerender because the multi-selection selectors it depends on would always map to return a new instance of an array. This would cascade down to each individual block because we assign a new dynamic ref function reference in the VisualBlockList component. This latter behavior is similarly non-ideal, but not as simple to resolve, so will be addressed separately.

Testing instructions:

Verify there are no regressions in the behavior of block multi-selection.

Ensure unit tests pass:

npm run test-unit

If you have React Developer Tools extension installed, toggle "Highlight Updates" and verify hovering only impacts relevant block (also layout, which will be debugged separately).

[codecov]

Codecov Report

Merging #3170 into master will increase coverage by 0.02%.
The diff coverage is 85.71%.

@@            Coverage Diff             @@
##           master    #3170      +/-   ##
==========================================
+ Coverage   31.61%   31.64%   +0.02%     
==========================================
  Files         217      217              
  Lines        6278     6282       +4     
  Branches     1116     1116              
==========================================
+ Hits         1985     1988       +3     
- Misses       3609     3610       +1     
  Partials      684      684

Impacted Files	Coverage Δ
editor/selectors.js	95.21% <85.71%> (-0.44%)	⬇️

---

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 b2e25cd...2f672df. Read the comment docs.

[mtias]

Testing looks good to me.