This will prevent string interpolation being evaluated during
release build. Especially useful in later commit where it becomes
more expensive.

… nodes

And rename previous type to KatexSpanNode, also while making it a
subtype of KatexNode.

… doc

The doc for message_details on this event says the field is present
if the flag is "read", with no mention of event_types. Looking
through zulip/zulip's Git history, it turns out it had been
mentioned in the past, but the mention was removed, in
zulip/zulip@48a1cf04d:

-  Present if `message` and `update_message_flags` are both present in
-  `event_types` and the `flag` is `read` and the `op` is `remove`.
+  Only present if the specified `flag` is `"read"`.

So the comment (and no-op `&& true`) isn't necessary; remove it.

We noticed this because the analyzer started flagging an info-level
`no_literal_bool_comparisons` here.

`flutter analyze` has started giving the following, which breaks CI:

   info • Unnecessary comparison to a boolean literal • lib/api/model/events.dart:1177:10 •
          no_literal_bool_comparisons
   info • Unnecessary comparison to a boolean literal • lib/model/channel.dart:158:13 •
          no_literal_bool_comparisons

We fixed the first one in the previous commit; this fixes the
second.

See discussion:
  https://chat.zulip.org/#narrow/channel/243-mobile-team/topic/CI.20on.20latest.20upstream/near/2228858

This lets us make it a bit more thorough, while not adding to the
complexity of reading the test case for the basic logic.

Fixes zulip#1743.

Potentially this would have helped us avoid the issue zulip#1743.

…itions

This test relies on background assumptions about how the "back" page
transition work which are true of the current default transitions on
Android (the default target platform of tests), but are broken by the
upcoming introduction of predictive back:
  flutter/flutter#165832 (comment)

For a quick unblocking of the upstream change, skip the test.

In principle there's no reason a test like this can't work with the
upcoming upstream behavior; it'll just need to work a little
differently.  But leave that for a follow-up change.

And update Flutter's supporting libraries to match.

This reverts commit 55d055f (zulip#1738).

That change was described as upgrading Flutter and its supporting
libraries, but it looks like it accidentally included other upgrades,
something like `flutter pub upgrade`.  It also didn't include the
CocoaPods updates for those upgrades; see zulip#1746.

So revert that change, to get back to a well-understood state.
Then we can retry the upgrade.

… soon

Thanks Komyyy for this idea, which I took from PR zulip#1363.

Co-authored-by: Komyyy <pol_tta@outlook.jp>

…ative

Servers can't yet start sending null without breaking clients.
Releasing this will start lowering the number of client installs
that would break, and eventually there will be few enough that the
breakage is acceptable; see discussion (same link as in comment):
  https://chat.zulip.org/#narrow/channel/378-api-design/topic/.60user_settings.2Etwenty_four_hour_time.60/near/2220696

See the doc:
  https://pub.dev/documentation/intl/latest/intl/DateFormat-class.html
> Formatting dates in the default 'en_US' format does not require
> any initialization.

(And we haven't been doing the described initialization for 'en_US'
or any other locale; it's asynchronous, and we have a better plan
for international formatting described in zulip#45, using ffi.)

We don't use these yet; coming up.

…tamps

Fixes-partly: zulip#1015

Fixes zulip#1015.