Claude review: drm/tidss: Add mechanism to detect DPI output

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

Patch Review

This patch adds `dpi_output` tracking through the pipeline. A few observations:

**Minor concern**: The `dpi_output` flag is stored both in `struct tidss_crtc` and redundantly written into `dispc->vp_data[].dpi_output` on every `dispc_vp_setup()` call. The value never changes at runtime, so writing it on every atomic flush is unnecessary overhead. It would be cleaner to set it once during init (e.g. in `dispc_init()` or via a dedicated setter), rather than threading it through every `dispc_vp_setup()` call. However, this is a minor design concern, not a bug.

**DPI detection heuristic**: The `tidss_is_bridge_internal()` function uses two heuristics - compatible string matching against `"ti,j721e-dsi"`, and the grandparent check for OLDI. This is somewhat fragile (if new internal bridges are added, the list must be updated), but the commit message acknowledges this and it is the best approach given the lack of a "DPI bridge" entity.

```c
+static const char * const tidss_internal_bridge_compatibles[] = {
+	"ti,j721e-dsi",
+};
```

This list may need future maintenance as new SoCs/bridges are added.

---
Generated by Claude Code Patch Reviewer