i18n: Annotate fragmented gettext locations with marker comments

[skkzsh]

Annotate fragmented gettext locations identified in #4074 with marker comments.

[AmirF194]

Clean annotation pass, and the important part is wired correctly: setup.cfg [extract_messages] has add_comments = TRANSLATORS:,dennis-ignore:, and every marker uses the {# TRANSLATORS: FIXME: fragment msgids (#4074) #} form, so pybabel strips the tag and writes the remainder into messages.pot as #. FIXME: .... The regenerated .pot in the PR confirms the comments are not inert, and they propagated into all the locale .po files so translators in every language see them. It is comments-only in the 8 templates (no non-marker changes, so no rendering impact), the one deletion is just the .pot timestamp header, and the marker keyword is consistent across all 23 files.

No blocking concerns. The lint-translations job already runs extract plus update-catalog and fails on any drift beyond the timestamp, so its passing is the verification that the catalog regenerates cleanly with these markers, and no separate functional test is really warranted for an annotation change.

Two optional thoughts: the note surfaces to translators as #. FIXME: fragment msgids (#4074), which reads as a developer FIXME. If the audience is translators, phrasing that says the sentence is split across strings and may read awkwardly (keeping the issue ref) would be more actionable for them. And if the goal is to cover all of #4074, the lint-template-i18n baseline counts 44 fragmented lines repo-wide while this adds 31 markers across 8 templates. One marker can cover a multi-fragment line so the counts need not match, but a quick note on whether this is the full set or a first batch would close the loop.