The expandable GPIO API requires the comparison of structured GPIO types. This means that inline functions must be used instead of direct comparisons. For the migration process, drivers must first be changed so that they use the inline comparison functions.
This reverts commit cf01c743a8.
Adding an unexplained delay seemed wrong in the first place, but it fixed
the display on the MCB2388.
Turns out the display was erroneously operating in 8-bit mode due to the
uninitialized flag register.
Why the delay helped here I don't know.
But with #12634 fixing this, this hack is not needed anymore.
The driver can only be used with either 4 or 8 bit modes. Checking if the 5th pin is set in the configuration is enough the determine if 8bit mode should be used or not
