Claude review: drm/radeon/radeon_connectors: remove radeon_connector_free_edid

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

Patch Review

**Approach:** Removes the `radeon_connector_free_edid()` helper and replaces all call sites with inline `drm_edid_free()` + NULL assignment. This is the right thing to do since `drm_edid_free()` is a standard API function.

**The transformation is mechanical and correct** — every call site is replaced with the same two-line pattern:

```c
drm_edid_free(radeon_connector->edid);
radeon_connector->edid = NULL;
```

**Minor observation:** The NULL assignment after `drm_edid_free()` is good practice but is technically unnecessary in the `radeon_connector_destroy()` path since the connector struct itself is about to be freed. Not worth changing, just noting.

**Ordering concern:** As mentioned above, this patch fixes the correctness problem created by patch 1 (where `kfree` was being called on `struct drm_edid *`). The series would be cleaner if either: (a) patch 1 updates `radeon_connector_free_edid` to use `drm_edid_free()` internally, or (b) the two patches are squashed. As submitted, bisecting to the state after only patch 1 produces broken code.

---

**Summary:** The API migration is correctly done and the final state after both patches is correct. The main concern is the bisectability issue: after patch 1 alone, `kfree()` is called on `const struct drm_edid *`, which is both a memory leak (inner edid not freed) and a const-correctness violation. Recommend either squashing or fixing the `radeon_connector_free_edid` body in patch 1.

---
Generated by Claude Code Patch Reviewer