Fix SAME interpreter to constrain to valid characters last

[ke4roh]

When a SAME message is received, it might be received with errors. There is substantial code to correct these messages, but it needs to be re-ordered. As it is, first constrains received characters to valid ones, then proceeds to find the nearest matches for words. Instead, it should 1. constrain for valid characters only as necessary to identify the shape (location of delimiters) of the message, 2. constrain to valid words (FIPS codes, valid message codes), and 3. constrain to valid characters last. This way, on step 2, it can accurately measure the number of bits changed to arrive at a valid result.