refactor(model): extract shared switch_model() from CLI and gateway handlers

[teknium1]

Summary

Phase 4 of the /model command overhaul — the refactor that unifies CLI and gateway model-switching logic.

What changed

Both handlers had ~50 lines of duplicated core logic (parsing, provider detection, credential resolution, model validation). Extracted into hermes_cli/model_switch.py:

Export	Purpose
ModelSwitchResult	Dataclass with all fields both handlers need
CustomAutoResult	Dataclass for bare /model custom results
switch_model()	Core pipeline: parse → detect → resolve → validate
switch_to_custom_provider()	Resolve endpoint + auto-detect model

The shared functions are pure (no I/O side effects). Each caller handles its own platform-specific concerns:

• CLI: sets self.model/provider/etc, calls save_config_value(), prints with ASCII emoticons
• Gateway: writes config.yaml directly, sets env vars, returns markdown

Impact

cli.py                     | 176 +++-------  (~80 lines of model handler, down from ~150)
gateway/run.py             | 179 +++-------  (~80 lines of model handler, down from ~150)
hermes_cli/model_switch.py | 234 ++++++++++  (new shared module)

Net: -244 lines from handlers, +234 in shared module. The handlers can't drift apart on core logic anymore.

Test plan

Full suite: 6090 passed, 0 failures. All existing /model tests pass unchanged — the refactor preserves exact behavior.