Fix dotted data-derived keys being split as nested paths in DotDict output

[Copilot]

DotDict.__setitem__ treats . as a nested-path separator, which corrupts output when field values are promoted to dict keys — as count does with count_key. A value like 'riko pt. 1' would silently become {'riko pt': {' 1': 1}}.

Changes

• riko/dotdict.py

  • DotDict.set_literal(key, value): stores a key verbatim via CaseInsensitiveDict.__setitem__, bypassing parse_key path-splitting entirely.
  • DotDict.__getitem__: before path-traversing a dotted key, checks key.lower() in self._store; if the full string was stored literally, returns it directly without splitting.

• riko/modules/count.py

  • parser builds grouped output items using set_literal instead of plain dicts, so any field value (URLs, version strings, sentences) used as a count_key round-trips without structural corruption.

Before / After

from riko.modules.count import pipe

stream = [{'title': 'riko pt. 1'}, {'title': 'riko pt. 2'}]
next(pipe(stream, conf={'count_key': 'title'}))
# Before: {'riko pt': {' 1': 1}}
# After:  {'riko pt. 1': 1}

Static developer-authored dotted paths in pipe configs are unaffected — set_literal is only called at sites that promote data values to keys.

• Fixes Dotted keys in data-derived output #60