Claude review: lib/fonts: Store font data as font_data_t; update consoles

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

Patch Review

This is the core type-system patch. **This patch has the most design concerns.**

The typedef:
```c
typedef const unsigned char font_data_t;
```

This is an unusual choice. `font_data_t *` becomes `const unsigned char *`, yet the code immediately casts away const when writing to allocated font data:
```c
memset((u8 *)new_data, 0, FONT_EXTRA_WORDS * sizeof(int));
...
memcpy((u8 *)new_data + i * h * pitch, data + i * vpitch * pitch, h * pitch);
```

A more conventional approach would be to use an opaque struct as the type (e.g., `struct font_data *`), which would prevent accidental direct pointer arithmetic while not requiring const-casting. The current approach means that `font_data_buf()` is essentially a no-op identity cast, and all the type safety comes from convention rather than the type system.

The `font_data_buf()` helper:
```c
static inline const unsigned char *font_data_buf(font_data_t *fd)
{
	return (const unsigned char *)fd;
}
```
This returns the same type as its argument. It exists as documentation that the caller wants the raw buffer, but adds no actual type safety.

In `fbcon_set_disp`:
```c
-	if (!vc->vc_font.data) {
+	if (!p->fontdata) {
```
This changes the semantics of the check — previously it checked `vc->vc_font.data`, now it checks `p->fontdata`. These could theoretically differ. The change looks intentional (the following code sets both from `t->fontdata`), but worth confirming the two are always in sync at this point.

---
Generated by Claude Code Patch Reviewer