Claude review: drm/amd/display: use GAMCOR for degamma private props in subsampled format

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

Patch Review

**Purpose**: When using subsampled video formats (4:2:0/4:2:2), the PRE_DEGAM hardware block (ROM curves) cannot be used because it would affect scaler linearity. This patch detects subsampled formats and passes that flag to `mod_color_calculate_degamma_params` so it computes distributed points for GAMCOR (RAM) instead.

**Correctness of the `map_user_ramp` repurposing**: Passing `is_subsampled_format` as the `map_user_ramp` parameter is semantically imprecise but functionally correct. When `true`, it prevents early returns at `color_gamma.c:1693-1696` that would keep `TF_TYPE_PREDEFINED`, forcing the function to fall through to `TF_TYPE_DISTRIBUTED_POINTS` computation. The `ramp=NULL` path is safe — all ramp accesses are guarded:

```c
// color_gamma.c:1718 — guards ramp dereference
if (map_user_ramp && ramp && ramp->type == GAMMA_RGB_256) {
```

```c
// color_gamma.c:1803 — same pattern
map_user_ramp && ramp && ramp->type == GAMMA_RGB_256,
```

**Whitespace issue**: Line 1457 of `amdgpu_dm_color.c` uses a tab + 7 spaces for the comment indentation, while surrounding code uses double-tabs. This is visible in the applied source:

```c
       } else {
	       /* Check if format requires ... */  ← tab + 7 spaces
		is_subsampled_format = ...             ← two tabs
```

The `} else {` at line 1456 also has space-based indentation, but that's a pre-existing issue.

**Error logging**: Adding `drm_err()` before returning `-ENOMEM` is a good improvement over the silent return in the original code.

**Nit**: The `is_subsampled_format` variable is declared at function scope but only used in the `else` branch. Declaring it inside the `else` block would be cleaner and make the scope explicit.

---
Generated by Claude Code Patch Reviewer