Claude review: drm: bridge: dw_hdmi: Rework HDP and RXSENSE interrupt handling

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

Patch Review

This is the largest and most impactful patch. It removes all RXSENSE handling, simplifies the IRQ handler to only care about HPD, and removes the `force`, `disabled`, `rxsense`, and `phy_mask` fields from `struct dw_hdmi`.

The simplified `dw_hdmi_phy_setup_hpd()` now sets HPD polarity based on current status:

```c
+	hdmi_modb(hdmi, hdmi_readb(hdmi, HDMI_PHY_STAT0) & HDMI_PHY_HPD ?
+		  0 : HDMI_PHY_HPD, HDMI_PHY_HPD, HDMI_PHY_POL0);
```

This avoids a spurious interrupt at probe/resume when a sink is already connected — the polarity is set to trigger only on the *next* change.

The `dw_hdmi_phy_update_hpd()` is emptied here (removed in patch 16). The `dw_hdmi_setup_rx_sense()` is also emptied (removed in patch 15). This two-step approach (empty then remove) is clean for bisection since external callers (meson, rockchip, sun4i, imx) need their references removed separately.

**Concern about `hdmi->force` removal**: Patch 10 adds `hdmi->force = connector->force;` in `dw_hdmi_connector_force()`, but this patch removes the `force` field from the struct. Looking carefully at the diff for this patch, the `force` assignment in `connector_force` is also removed here:

```c
 	mutex_lock(&hdmi->mutex);
-	hdmi->force = connector->force;
 	hdmi->last_connector_result = connector->status;
```

So this is correct — the field and all its uses are removed atomically in this patch. My earlier concern was unfounded after tracing the full diff. **No Reviewed-by yet.**

---
Generated by Claude Code Patch Reviewer