Claude review: drm/msm/dp: Move link training to atomic_enable()

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

Patch Review

Moves `msm_dp_ctrl_on_link()` from `msm_dp_display_process_hpd_high()` to `msm_dp_bridge_atomic_enable()`.

**Issues:**

1. The `// TODO: schedule drm_connector_set_link_status_property()` comment should be addressed or at least converted to a proper C-style comment:
```c
+	// TODO: schedule drm_connector_set_link_status_property()
```
The kernel coding style requires `/* */` comments. Also, if link training fails, the function continues into `msm_dp_display_enable()` anyway, which seems wrong — a failed link train should probably bail out.

2. The commit message says it also changes hpd_notify() behavior ("call the plug/unplug handlers directly in hpd_notify() instead of queueing them"), but there's no change to hpd_notify() in this patch. The description seems to be a leftover from an earlier version.

---
Generated by Claude Code Patch Reviewer