qcom: clock-alpha-pll: Correct types to account for error values

Using an unsigned 32 bit integer to contain error
values is invalid, since error checking macros
such as IS_ERR_VALUE may promote these variables
to 64 bits before comparing with the error values.

Change-Id: I67834897fc7a664e5ca68cb8a61a3172a6e36853
Signed-off-by: Vikram Mulukutla <markivx@codeaurora.org>