Claude review: mfd: lm3533: Convert to use OF bindings

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

Patch Review

This is the largest patch. The overall conversion from pdata to DT is well done — dynamic `mfd_cell` allocation from child nodes, parsing DT properties, removing all pdata structs.

**Bug: DT property name mismatch for ALS resistor.**

The binding schema (patch 1) defines the property as `ti,resistor-ohms`, but the driver reads:

```c
		ret = device_property_read_u32(dev, "ti,resistor-value-ohms",
					       &als->r_select);
```

This should be `"ti,resistor-ohms"` to match the binding.

**Typo in error message:**

```c
		return dev_err_probe(dev, ret,
				     "failed to ger resistor value\n");
```

"ger" should be "get".

**Minor observations:**

- The resistor ohms-to-register conversion `DIV_ROUND_UP(2 * MICRO, 10 * als->r_select)` is correct: for 1575 ohms → register 127, for 200000 ohms → register 1.

- `led->cdev.default_trigger = "none"` is unconventional; the standard approach is to leave it NULL when no trigger is specified. The subsequent `device_property_read_string` will overwrite it only if the DT property exists.

- Moving `LM3533_MAX_CURRENT_*` defines from `lm3533-ctrlbank.c` to the shared header is necessary since both LED and backlight drivers now clamp current values before calling `lm3533_ctrlbank_set_max_current`.

- The `boost_ovp` and `boost_freq` values are not initialized before the `clamp()` call. If the DT property is missing, `device_property_read_u32` leaves the variable uninitialized. Since `boost_ovp` and `boost_freq` are struct members (zero-initialized via devm_kzalloc), the default value 0 will be clamped to the minimum. This works but could be more explicit.

---
Generated by Claude Code Patch Reviewer