Claude review: video: leds: backlight: lm3533: Support getting LED sources from DT

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

Patch Review

**Bug: Wide mask clobbers other banks' LED source configurations.**

In `lm3533_bl_setup`:
```c
	ret = regmap_update_bits(bl->lm3533->regmap, LM3533_REG_OUTPUT_CONF1,
				 OUTPUT_CONF1_MASK, output_cfg_val);
```

`OUTPUT_CONF1_MASK = GENMASK(1, 0) = 0x03` covers BOTH HVLED1 and HVLED2 fields. If backlight bank B only configures HVLED1 (led-sources = <0>), the mask still writes bit 1 as 0, resetting HVLED2's assignment to bank A — even if bank A had previously configured HVLED2 differently. The same issue exists on the LED side:

```c
	ret = regmap_update_bits(led->lm3533->regmap, LM3533_REG_OUTPUT_CONF1,
				 OUTPUT_CONF1_MASK,
				 output_cfg_val << OUTPUT_CONF1_SHIFT);
```

`OUTPUT_CONF1_MASK = GENMASK(7, 2)` covers all three LVLED1/2/3 fields. Configuring one LVLED clobbers the other two. The original `lm3533_set_hvled_config` and `lm3533_set_lvled_config` correctly computed per-LED masks to avoid this. The fix is to build the mask dynamically alongside the value:

```c
	for (i = 0; i < bl->num_leds; i++) {
		if (bl->led_strings[i] >= LM3533_HVCTRLBANK_COUNT)
			continue;
		output_cfg_mask |= LM3533_BL_ID_MASK << bl->led_strings[i];
		output_cfg_val |= id << bl->led_strings[i];
	}
	ret = regmap_update_bits(..., output_cfg_mask, output_cfg_val);
```

**Possible undefined symbol:** `LM3533_LVCTRLBANK_MAX` is used in `leds-lm3533.c` for the `leds[]` array size and bounds check, but I don't see its definition in any patch diff. Verify this compiles — it may need to be added to the header or as a local define.

---
Generated by Claude Code Patch Reviewer