By what factor is a keyspace reduced by using a pre defined amount of ascii character types vs filtering input data directly through tr?
Is it negligible?
pseudo-code examples (assuming each group of 4 bytes are filtered from a well stretched KDF):
$(4 uppercase | 4 lowercase | 4 ascii (non-alphanumeric) | 4 digits) -> shuffle chars using fixed binary seed
$(1000 byte binary input) -> filter non ascii bytes -> trim to 16 bytes
A 16 byte keyspace is a 16 byte keyspace, regardless of how you enter it
>>61850443
a000000000000001