Update documentation for LM(layer, mod) (#19284)

* Update documentation for LM(layer, mod)

* Move LM table to caveats section

docs/feature_layers.md

@@ -10,7 +10,7 @@ These functions allow you to activate layers in various ways. Note that layers a
 
 * `DF(layer)` - switches the default layer. The default layer is the always-active base layer that other layers stack on top of. See below for more about the default layer. This might be used to switch from QWERTY to Dvorak layout. (Note that this is a temporary switch that only persists until the keyboard loses power. To modify the default layer in a persistent way requires deeper customization, such as calling the `set_single_persistent_default_layer` function inside of [process_record_user](custom_quantum_functions.md#programming-the-behavior-of-any-keycode).)
 * `MO(layer)` - momentarily activates *layer*. As soon as you let go of the key, the layer is deactivated.
-* `LM(layer, mod)` - Momentarily activates *layer* (like `MO`), but with modifier(s) *mod* active. Only supports layers 0-15 and the left modifiers: `MOD_LCTL`, `MOD_LSFT`, `MOD_LALT`, `MOD_LGUI` (note the use of `MOD_` constants instead of `KC_`). These modifiers can be combined using bitwise OR, e.g. `LM(_RAISE, MOD_LCTL | MOD_LALT)`.
+* `LM(layer, mod)` - Momentarily activates *layer* (like `MO`), but with modifier(s) *mod* active. Only supports layers 0-15. The modifiers this keycode accept are prefixed with `MOD_`, not `KC_`. These modifiers can be combined using bitwise OR, e.g. `LM(_RAISE, MOD_LCTL | MOD_LALT)`.
 * `LT(layer, kc)` - momentarily activates *layer* when held, and sends *kc* when tapped. Only supports layers 0-15.
 * `OSL(layer)` - momentarily activates *layer* until the next key is pressed. See [One Shot Keys](one_shot_keys.md) for details and additional functionality.
 * `TG(layer)` - toggles *layer*, activating it if it's inactive and vice versa
@@ -21,6 +21,12 @@ These functions allow you to activate layers in various ways. Note that layers a
 
 Currently, the `layer` argument of `LT()` is limited to layers 0-15, and the `kc` argument to the [Basic Keycode set](keycodes_basic.md), meaning you can't use keycodes like `LCTL()`, `KC_TILD`, or anything greater than `0xFF`. This is because QMK uses 16-bit keycodes, of which 4 bits are used for the function identifier and 4 bits for the layer, leaving only 8 bits for the keycode.
 
+For a similar reason, the `layer` argument of `LM()` is also limited to layers 0-15 and the `mod` argument must fit within 5 bits. As a consequence, although left and right modifiers are supported by `LM()`, it is impossible to mix and match left and right modifiers. Specifying at least one right-hand modifier in a combination such as `MOD_RALT|MOD_LSFT` will convert *all* the listed modifiers to their right-hand counterpart. So, using the aforementionned mod-mask will actually send <kbd>Right Alt</kbd>+<kbd>Right Shift</kbd>. Make sure to use the `MOD_xxx` constants over alternative ways of specifying modifiers when defining your layer-mod key.
+
+| `LM(1,KC_LSFT)` | `LM(1,MOD_MASK_SHIFT)` | `LM(1,MOD_BIT(KC_LSFT))` | `LM(1,MOD_LSFT)` |
+|:---------------:|:----------------------:|:------------------------:|:----------------:|
+|       ❌        |          ❌            |           ❌             |        ✅        |
+
 Expanding this would be complicated, at best. Moving to a 32-bit keycode would solve a lot of this, but would double the amount of space that the keymap matrix uses. And it could potentially cause issues, too. If you need to apply modifiers to your tapped keycode, [Tap Dance](feature_tap_dance.md#example-5-using-tap-dance-for-advanced-mod-tap-and-layer-tap-keys) can be used to accomplish this.
 
 ## Working with Layers :id=working-with-layers