Issue #3616958: Say only what the code does now, and keep Orchestra in one place

Three leftovers from [#3616250], in one MR.

Comments describing a design that is gone

Sixteen in shipped code, reworded to state the reason in the present tense. The reasons are kept, because what the alternative costs is why the code is shaped the way it is; only the narration of a former shape goes.

Comments in TESTS are deliberately left: there "it used to be X" names the defect the test pins, which is the test's whole purpose. There are 21 of those, not the five the issue summary guessed at; the summary is corrected.

Dead entries in the French translation

Three msgids whose source string is nowhere in the tree, two of them a placement field that was removed. Checked with apostrophes normalized on both sides: a YAML single-quoted scalar writes order''s, and a literal search called three LIVE strings dead before that was accounted for.

Every Orchestra integration under yoyaku_orchestra

yoyaku_payment_orchestra, yoyaku_cart_orchestra and yoyaku_ticket_orchestra become yoyaku_orchestra_payment, yoyaku_orchestra_cart and yoyaku_orchestra_ticket, beside yoyaku_orchestra_order and yoyaku_orchestra_manager. Machine names, namespaces, service ids, config schema, the example's dependency, the human-readable names and the docs all follow. Lifting Orchestra out later is one directory rather than four, and no Orchestra-dependent module is left stranded under a module that stays.

Pre-1.0, so the renames need no upgrade path: a site reinstalls the submodules under their new names.

Verified

phpcs (Drupal and DrupalPractice), phpstan, cspell and the CI-faithful eslint gate all clean. Kernel suites green: the whole yoyaku_orchestra tree including the three moved bridges (23 classes), plus yoyaku_cart, yoyaku_ticket, yoyaku_payment and yoyaku_placement (95 classes). The moved ticket bridge's unit test passes against a root where the module actually lives at its new path. ModuleBoundariesTest still green at 31 modules.