docs: copy-edit guides and fix demo-notebook narrative/code inconsistencies

[thodson-usgs]

What

A documentation copy-edit pass plus fixes for narrative/code inconsistencies in the demo notebooks. All 20 docs notebooks were executed end-to-end against the live USGS Water Data API; every fix below was verified against live output before editing, and the edited notebooks were re-run to confirm they pass.

Notebook fixes (demos/)

One notebook failed to execute before this PR, the rest had narrative that contradicted their own output:

• DiscreteSamples (was crashing) — get_codes returns a (df, md) tuple, but five cells indexed the tuple with a column list (TypeError), and the prose claimed it "returns a plain DataFrame". Unpack the tuple; correct the claim.
• SiteInfo — state_code="UT" returned an empty frame under an "all locations in a state" heading. state_code is a two-digit ANSI code → use "49" (Utah).
• UnitValues — two notes said returned timestamps are "in local time"; they are tz-aware UTC. Removed a dead duplicate of Example 5.
• Samples — "181 fields" → the default profile returns 187 columns; replaced six references to non-existent *_lookup() helpers with the real get_codes(code_service=...).
• GroundwaterLevels — stale comment claimed partial dates "show up as NaT" in the index; the index is a plain RangeIndex and dates live in a normalized UTC time column. Relabeled a y-axis that mixed depth-below-surface with NGVD29/NAVD88 elevations.
• Introduction — get_combined_metadata joins monitoring-location + time-series metadata, not "field-measurement metadata".
• R vignette — nwis.get_water_use() is defunct (raises NameError); noted instead of shown as runnable.
• SiteInventory — Example 3 duplicated Example 2 verbatim → repurposed as a skip_geometry=True demo.
• peak_streamflow_trends — the live-data migration (ad4e980f) removed the CSV-load cell but left narrative cells describing it; rewrote them to match the live final_df, and corrected a chunker comment (RI's 350 gages return in a single request — verified).
• Disambiguated two identical get_field_measurements() notebook titles (Surface-Water vs Groundwater-Level).

Prose docs

• installing.rst: fixed the conda command (conda install -c conda-forge dataretrieval).
• contributing.rst: version is derived from Git tags via setuptools_scm, so the "edit setup.py / conf.py" steps are obsolete → tag-based release note; Python "3.6, 3.7, 3.8" → "3.9 and later".
• CONTRIBUTING.md: stale USGS-python/dataretrieval URLs → DOI-USGS/dataretrieval-python; blob/master → blob/main; Python versions; "interace" typo.

Cleanup

• Removed docs/source/examples/datasets/peak_discharge_trends.csv (535 KB) — leftover from the ad4e980f live-data migration, referenced nowhere in the repo.

Verification

• 20/20 docs notebooks execute against the live API (1 previously failed → now passes); all edited notebooks re-executed clean.
• Embedded .rst doctests (readme_examples, siteinfo_examples, timeconventions) verified exactly against live output.
• pre-commit (ruff, nbstripout, eof) passes on all changed files.

🤖 Generated with Claude Code