This binding currently duplicates the permitted clock-names in various
places, and when adding more compatibles, clock-names will have to be
duplicated even more.

The reason is:
1) subschemas (-if: ...), still have to match the top-level:
       pattern: '^clk_uart_baud[0-3]$'
2) there is one compatible that doesn't follow sequential numbering for
   the clock names (samsung,s3c6400-uart)
3) when limiting the number of clock-names, we also want to enforce
   sequential names
Because of 1) and 2), the patterns can not simply be changed to
constant strings, and later overridden in a different subschema (for
samsung,s3c6400-uart only).

Since we can't populate the top-level clock-names based on the
compatible, and because when limiting the number of items we generally
want sequential numbers and not a pattern, move the permitted strings
into a subschema of its own and populate it based on the compatible:
    * 'uart clk_uart_baud2 clk_uart_baud3' for the one outlier
    * 'uart clk_uart_baud0..3' for everything else

This way we can avoid having to duplicate the permitted names
everywhere.

While at it, add blank lines as per the universal style, which is to
have blank lines between properties, except where they are booleans.

Also add another example using a compatible that uses the default
clock-names scheme, as opposed to the existing example that uses
samsung,s3c6400-uart's non-default clock-names. This allows testing
both versions of the clock-names property when running
dt_binding_check.
Reviewed-by: Rob Herring (Arm) <robh@kernel.org>
Signed-off-by: André Draszik <andre.draszik@linaro.org>
Link: https://lore.kernel.org/r/20240725-gs101-uart-binding-v5-1-e237be8253a9@linaro.orgSigned-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>