fix(waterdata): keep sublocation_identifier from shifting positional arguments

[thodson-usgs]

Follow-ups from a review of #436 after it merged. #436 is not in a release yet (v1.4.0 predates it), so moving the parameter breaks no released caller.

• Signature order. fix(waterdata): accept field-measurements sublocation_identifier #436 inserted sublocation_identifier mid-signature in get_field_measurements (before skip_geometry) and get_field_measurements_metadata (before properties). Every later positional argument shifted, so a positional skip_geometry or properties would have been sent as a sublocation filter. The parameter now comes last, before **queryables. get_continuous's method_category (fix(waterdata): accept the continuous method_category queryable #423) has the same problem but shipped in v1.4.0, so it is left alone.
• Test. The shared fixture's site (USGS-05427718) records no sublocation, so the test only checked that a hand-added null column exists. It now uses real features captured from USGS-05435943, one each for Backup, HWM and Primary (new tests/data/waterdata_sublocation_fixtures.json), and asserts those values.
• Docs. get_time_series_metadata and get_combined_metadata listed sublocation_identifier with no description; it is described now, with free-text values observed there (YSI DOWN, 30 CM). The field-measurement examples now use observed values; UPSTREAM was not among them.
• NEWS. The entry is re-dated to its land date (10/07), its hard wraps are removed, and it notes that the field already worked through **queryables.

Not done: an offline test comparing each getter's parameters with tests/data/waterdata_queryables.json. Queryables are filters and not all are meant to be named parameters (monitoring-locations has about 240), so the existing /schema columns check from #424 is the better place for this.

Overlap with #446

#446 rewrites the list-case assertion in test_sublocation_identifier_is_sent_as_a_filter, from a comma join to a CQL2 in. This PR changes the adjacent lines, so whichever PR merges second needs a trivial rebase.

Verification

• 1,254 offline tests pass, coverage 99.02%; ruff, mypy, xenon, complexipy and import-linter are clean.
• The live queryables snapshot test passes (11/11).
• Live: get_field_measurements(monitoring_location_id="USGS-05435943", sublocation_identifier="HWM") returns 2 rows, and get_field_measurements_metadata(..., sublocation_identifier="Backup") returns 1.

🤖 Generated with Claude Code

https://claude.ai/code/session_01HBt3Ee7eneKtdKQWq2T3YK