Claude review: drm/panel: novatek-nt36672a: Inline panel init sequences

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

Patch Review

**Correctness: Good.** The conversion from table-driven to inline is faithful. All loop ranges are correct, all values match, and the `0x1` vs `0x01` difference in the original `0xFB` register write is a non-issue (identical in C).

**Minor nit — unnecessary NULL checks on function pointers:**

```c
if (pinfo->desc->send_deinit_cmds)
    pinfo->desc->send_deinit_cmds(&dsi_ctx);
```

```c
if (pinfo->desc->send_init_cmds_1)
    pinfo->desc->send_init_cmds_1(&dsi_ctx);
```

```c
if (pinfo->desc->send_init_cmds_2)
    pinfo->desc->send_init_cmds_2(&dsi_ctx);
```

The original code passed `off_cmds` / `on_cmds_1` / `on_cmds_2` unconditionally (a NULL pointer with `num=0` would just skip the loop). Since there is currently only one panel variant (`tianma_fhd_video_panel_desc`) and it sets all three function pointers, these NULL checks are strictly unnecessary today. However, they are **reasonable defensive coding** for when future panel variants are added that might not need all three init phases — so this is a fine design choice. No action needed.

**`accum_err` propagation:** The function pointer calls don't check or propagate `dsi_ctx.accum_err` themselves, but they don't need to — the `_multi` helpers internally short-circuit on `accum_err != 0`, and the callers (`prepare`/`unprepare`) already handle or reset errors at the appropriate points. This matches the existing behavior.

**Style:** Hex values correctly lowercased per v3 changelog. Comments preserved where meaningful (skin enhancement mode, UI/STILL/MOVING mode, dimming, resolution).

**No functional change** beyond the refactoring itself — the same bytes get written to the same registers in the same order.

Reviewed-by worthy as-is.

---
Generated by Claude Code Patch Reviewer