Handle number-less DateTimeFormat Hebrew month

[justingrant]

The ICU formats for the Hebrew calendar no longer output a month number, only the name. This PR works around this change. See https://bugzilla.mozilla.org/show_bug.cgi?id=1751833 for more context.

Fixes #2015.

[codecov]

Codecov Report

Merging #2034 (1b1657e) into main (ad37d6e) will decrease coverage by 0.01%.
The diff coverage is 87.50%.

@@            Coverage Diff             @@
##             main    #2034      +/-   ##
==========================================
- Coverage   94.44%   94.43%   -0.02%     
==========================================
  Files          19       19              
  Lines       10989    11002      +13     
  Branches     1596     1597       +1     
==========================================
+ Hits        10379    10390      +11     
- Misses        590      592       +2     
  Partials       20       20              

Flag	Coverage Δ
test262	81.15% <25.00%> (-0.04%)	⬇️
tests	88.73% <75.00%> (-0.02%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
polyfill/lib/calendar.mjs	94.25% <87.50%> (-0.07%)	⬇️

---

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 ad37d6e...1b1657e. Read the comment docs.

[Ms2ger]

Can we test this?

[justingrant]

Can we test this?

If Node's ICU data is updated to include the problematic change (the same one that Chrome and FF have already picked up) then any property access using the Hebrew calendar will trigger this bug. So it will be tested by CI Demitasse tests (in intl.mjs) as soon as CI is updated to use a version of Node with the latest ICU data.

In the meantime, I manually verified it in latest Chrome and latest FF and Chrome. Before the fix in this PR, the following code will throw. After the fix, it won't.

Temporal.PlainDate.from('2020-01-01[u-ca=hebrew]').day;

Your comment made me realize that Node now has v17, so we should probably be testing it in CI. I'll open a new issue for that.

[justingrant]

BTW, I confirmed that the current (pre-PR) code fails under Node 17 tests. See https://github.com/tc39/proposal-temporal/runs/5069846827?check_suite_focus=true

Then I rebased the Node 17 PR (#2039) on top of this PR and verified that Node 17 passes.