Bump ecmarkup and enable "user code" annotation

[bakkot]

cc @gibson042

See tc39/ecma262#2548 for a description of the annotation. The relevant information is now propagated in the biblio, so it can be used in proposals. In ecmarkup the annotation off by default because it may require manual annotation to be correct, but I've enabled it in this PR: I don't think any sites in this PR need annotation - there's no use of [[Get]] and friends directly, no weird stuff with generators, etc. (I didn't review extremely carefully, though.)

Also

• makes use of add redefinition support to structured headers ecmarkup#428 to get rid of the ​ hack.
• marks a couple of uses of the phrase "time value" as not being in the more precise ecma262 sense of the term.

[codecov]

Codecov Report

Merging #2125 (bd9529e) into main (1f3fba8) will not change coverage.
The diff coverage is n/a.

@@           Coverage Diff           @@
##             main    #2125   +/-   ##
=======================================
  Coverage   92.19%   92.19%           
=======================================
  Files          19       19           
  Lines       10928    10928           
  Branches     1571     1571           
=======================================
  Hits        10075    10075           
  Misses        835      835           
  Partials       18       18           

Flag	Coverage Δ
test262	82.60% <ø> (ø)
tests	84.10% <ø> (ø)

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

---

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 1f3fba8...bd9529e. Read the comment docs.

[bakkot]

I think the codecov bot is broken.

[ptomato]

I think the codecov bot is broken.

No, just really slow to execute test262 tests 😛

[ptomato]

Thanks!

This might be obvious if I look at the documentation for the user code feature in ecmarkup, but should we be specifically annotating when a call site marked "UC" cannot call into user code? e.g. CreateTemporalDate does not call into user code when newTarget is absent, and there are places in the spec text where it is called without a newTarget.

[bakkot]

This might be obvious if I look at the documentation for the user code feature in ecmarkup, but should we be specifically annotating when a call site marked "UC" cannot call into user code? e.g. CreateTemporalDate does not call into user code when newTarget is absent, and there are places in the spec text where it is called without a newTarget.

The heuristic ecmarkup uses is that if you are calling something which normally triggers user code, but are using ! to assert that it can't throw in this particular use, that probably means it's not calling user code (otherwise it would have been able to throw). This heuristic is almost perfectly accurate on the main 262 spec, but I don't how how well it works here. At a skim, it looks roughly right for CreateTemporalDate except that there are a bunch of calls to CreateTemporalDate which are ?-prefixed but also don't pass newTarget, which seems wrong to me.

If you are calling something which can sometimes invoke user code in such a way that it cannot invoke user code in this particular usage, but can still throw, you will need to manually suppress the effect, yes.