docs(15.9): correct eight places where the docs disagree with the product

[marevol]

Summary

A release test of 15.9 found eight places where the documentation describes behaviour that Fess does not have. The same text is in 15.8. This PR fixes the 15.9 tree in all seven languages.

Page	Was	Now
api/admin/api-admin-backup.rst	GET /api/admin/backup/file/fess_config.bulk	A path whose last segment contains a dot and has no trailing / is treated as a static file request and never reaches the API, so it returns 404. The page now says to end such ids with /, and both curl examples do.
user/search-sort.rst	An unsupported field or order in sort: makes the search fail	In the query, an unsupported sort: does not raise an error. The query is escaped and searched again, so the results are not sorted and usually nothing is found. Only the sort request parameter returns 400.
config/search-advanced.rst	query.additional.api.response.fields adds fields to the API response	That key only extends the search API's allow-list. The field must also be fetched through query.additional.response.fields. There is now a note and an example that sets both.
api/api-search.rst, api/admin/api-admin-searchlist.rst	record_count_relation is eq / gte	EQUAL_TO / GREATER_THAN_OR_EQUAL_TO. These are the values the API returns and the bundled OpenAPI definition lists.
admin/log-guide.rst	search.log	searchlog.log
config/admin-logging.rst	Rotation by size, with at most 10 generations	The bundled log4j2.xml also deletes compressed files by age: 90 days for fess.log, fess-llm.log and searchlog.log, and 360 days for audit.log. The page now lists these periods and the properties that set them, and says the 10-file limit applies per date. The French page also said 10 MB where the limit is 100 MB.
admin/stopwords-guide.rst	Copied from the mapping guide ("source" / "target" characters)	The page describes stopwords and the single word field on the screen. It also gives the download and upload format: one word per line, and lines starting with # are comments.
admin/stopwords-guide.rst	No note on scope	A new note says each stopwords file is used only by the analyzer for its own language. A word added only to en/stopwords.txt is removed from content and title. It can still match through content_ja of documents detected as Japanese, because language-specific fields are added to the query based on the request language.

Verification

• Checked each statement against the Fess source: the backup API action and LastaFlute request routing, SearchHelper / TermQueryCommand / QueryStringBuilder, QueryFieldConfig and the v2 search handlers, log4j2.xml, the stopwords admin form, JSP and dictionary file parser, and the analyzers and dynamic templates in the bundled index settings.
• Parsed every changed file with docutils (doctitle_xform=False). Each file still has exactly one top-level section, and none has a new warning.

Follow-up: how the response-field keys combine

A second commit covers three more places, all about fields that are fetched from the index and fields the API is allowed to return.

Page	Was	Now
config/properties.rst	Short descriptions of query.additional.response.fields, query.additional.api.response.fields and query.additional.scroll.response.fields	Regenerated from the rewritten comments in codelibs/fess#3490. They explain that one set of keys fetches a field and the other lets the API return it. Only these three rows change, and the per-language catalogues get the new msgids. Like every other entry in those catalogues, they are not translated yet. Merge this PR after codelibs/fess#3490.
api/admin/api-admin-backup.rst	fess.json / doc.json are scrolled like a .bulk id	The download returns the mapping definition file itself (fess_indices/fess.json, fess_indices/fess/doc.json) as application/octet-stream. The ID table now has a row for them. The note now says that the API only downloads and that files are restored from the Backup page in the admin screen. fess.json is also listed among the ids that need a trailing /.
config/search-scroll.rst	query.additional.scroll.response.fields=content makes /api/v2/documents/all return content	The scroll API also filters every document through the API response allow-list, which does not include content by default. The example now sets both query.additional.scroll.response.fields=content and query.additional.api.response.fields=content, and both notes explain why both are needed.

Verified against ApiAdminBackupAction#get$file, AdminBackupAction (upload and mapping-file paths), ScrollSearchHandler#filterDoc, QueryFieldConfig (default scroll fields and API allow-list) and SearchHelper. tools/gen_properties_doc.py --check and tools/test_gen_properties_doc.py pass. None of the changed files gets a new docutils warning.