msglist test: Add test for messagesCloseInTime, and update a related test

[rajveermalviya]

Fixes: #1825

[gnprice]

Thanks! Generally this revision looks good; just small comments.

Given your timezone you're probably asleep at this hour, and I'll be off on vacation before you start your next workday. So @chrisbobbe please pick up the reviews from here, and go ahead and merge this PR when you consider it ready.

[gnprice]

nit:

msglist [test]: Avoid specifying explicit message-IDs in `eg.{*}Message()`

"msglist test", rather than "msglist [test]"

[gnprice]

nit: these two commits should get squashed together:

50c8123 msglist test [nfc]: Refactor forcedShowSender evaluation in checkInvariants
b2e9f4e msglist test [nfc]: Rename forcedShowSender to showSender in checkInvariants

for the same reason that motivated the rename (#1815 (comment)): otherwise the intermediate state is confusing in that the name doesn't match the meaning

[gnprice]

This is a nice logical sequence. One more case will complete the sequence:

Suggested change

	doTest('2022-01-01 00:30:00', time, false);
	doTest('2021-02-01 00:30:00', time, false);
	doTest('2021-01-02 00:30:00', time, false);
	doTest('2022-01-01 00:30:00', time, false);
	doTest('2021-02-01 00:30:00', time, false);
	doTest('2021-01-02 00:30:00', time, false);
	doTest('2021-01-01 01:30:00', time, false);

[gnprice]

Let's also have this set of cases exercise the normal situation where the first time is earlier than the second one, instead of the unusual case where it's later. That way these will be exercising cases we expect to happen regularly, rather than cases that can only happen if something's gone wrong somewhere.

(The current implementation will clearly behave the same either way; but that's not something that's evident from looking at the test and the name of the function, and not something that necessarily has to stay true in the future.)

[rajveermalviya]

Thanks for the review @gnprice!

@chrisbobbe, pushed an update, PTAL.

[chrisbobbe]

Thanks, LGTM! Merged.