Apply _normalize_str_iterable to every multi-value string parameter

Extracts the type-and-iterable normalization out of
_check_monitoring_location_id into a reusable helper, then wires it
into every public waterdata getter at every multi-value string
parameter. Closes the gap noted in PR review: previously only
`monitoring_location_id` accepted pd.Series / np.ndarray / tuple;
other params (parameter_code, statistic_id, state_name, etc.) still
silently str-serialized non-list iterables into the request.

Changes:

- dataretrieval/waterdata/utils.py: new `_normalize_str_iterable(value,
  param_name)` does the None / str / non-Mapping-Iterable dispatch and
  per-element type check. `_check_monitoring_location_id` now wraps it
  and adds the AGENCY-ID hyphen check, which remains
  monitoring_location_id-specific.

- dataretrieval/waterdata/api.py: 153 normalization calls inserted
  across 11 public functions (get_daily, get_continuous,
  get_monitoring_locations, get_time_series_metadata,
  get_latest_continuous, get_latest_daily, get_field_measurements,
  get_samples, get_stats_por, get_stats_date_range, get_channel). 170
  type annotations widened from `str | list[str] | None` / `list[str]
  | None` to `str | Iterable[str] | None`. Added Iterable import and
  `_normalize_str_iterable` to imports.

  Excluded (kept as-is): time-range params (time, last_modified,
  begin, end, begin_utc, end_utc, datetime) — these have special
  semantics in _format_api_dates (single-string or two-element range).
  Out of scope; users who need iterable support there can `.tolist()`.

- tests/waterdata_test.py: new TestNormalizeStrIterable class with 10
  tests covering str / list / tuple / pd.Series / pd.Index /
  np.ndarray acceptance, plus int / dict rejection, plus an
  integration check (mock.patch on get_ogc_data) confirming that
  passing pd.Series for parameter_code arrives at the inner call as a
  list, not a stringified Series.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

Author: thodson-usgs