Claude review: drm/bridge: ite-it66121: Move logic .mode_set setup to .atomic_enable

[Claude Code Review Bot <claude-review@example.com>]

Patch Review

**Good:**
- Correctly moves from the deprecated `.mode_set` to `.atomic_enable`.
- The new `it66121_set_mode` properly extracts the mode from the atomic state through the connector → CRTC → adjusted_mode chain.

**Issue — ordering of `it66121_set_mode` vs `drm_atomic_helper_connector_hdmi_update_infoframes`:**

In the resulting `atomic_enable`:

```c
	drm_atomic_helper_connector_hdmi_update_infoframes(ctx->connector, state);

	it66121_set_mode(ctx, state);
	it66121_set_mute(ctx, false);
```

The infoframes are written *before* the hardware is configured (TX mode, clock bank, AFE, etc.). In the original code, the infoframe write happened in `mode_set` which runs before `atomic_enable`, so the ordering was the same. But logically, writing infoframes before the TX mode is set seems backwards — the hardware should probably be configured first, then infoframes sent. Looking at the register programming sequence (power off TX clock → configure input → configure AFE → power on TX clock → set HDMI mode), infoframes should come *after* the TX path is set up. Consider swapping the order so `it66121_set_mode` runs before the infoframe update. However, if this matches the tested behavior on real hardware, it may be intentional.

**Observation — no error propagation from `it66121_set_mode`:**

The function silently swallows errors via `goto unlock`. This isn't new (the old `mode_set` did the same), but now that it's called from `atomic_enable`, it might be worth at least a `dev_err()` on failure so issues are visible in dmesg.

---
Generated by Claude Code Patch Reviewer