docs: fix noun-stack over 3 modifiers in metadata docstring

[thodson-usgs]

Summary

Repo-wide review for noun stacks deeper than 3 modifiers before a head noun (per the style rule: cap at three modifiers, convert the rest to a prepositional phrase).

Scanned prose across README.md, AGENTS.md, CONTEXT.md, CONTRIBUTING.md, NEWS.md, docs/source/**, and every docstring/comment in dataretrieval/. The corpus is deliberately edited prose (per the domain-modeling/writing-for-agents conventions already in place) and almost entirely clean. Found one genuine violation:

• dataretrieval/waterdata/metadata.py (get_monitoring_locations docstring, aquifer_code parameter): "the USGS water resources data base" — four modifiers (USGS, water, resources, data) stacked before the head noun base. Rewritten as "the USGS database of water resources", converting one modifier into a prepositional phrase. Also fixes the stray "data base" → "database" spelling inconsistency with the rest of the codebase.

Not changed (reviewed and kept as-is)

• "the new Water Data STAC catalog" (NEWS.md) and its recurrences in dataretrieval/waterdata/ratings.py, dataretrieval/rdb.py, and demos: technically 4 modifiers before "catalog," but Water Data is a frozen core-term/service name (CONTEXT.md) and STAC is a fixed external standard's acronym — this is one consistently-used compound proper noun, not an arbitrary stack, and it appears identically across multiple files. Rewriting it in one spot would create inconsistency without a clear readability gain.
• The equivalent phrasing in R/dataRetrieval/man/*.Rd and R/dataRetrieval/vignettes/*.Rmd is out of scope — AGENTS.md explicitly marks R/ as untracked local scratch, not part of this repo.

Testing

• ruff format --check / ruff check pass on the changed file.
• Pre-commit hooks (ruff, mypy, xenon, complexipy, import-linter) all passed on the commit.
• Docstring-only change; no behavior, no tests needed.