Draft: ASCII85 rework

a bit of a tangent, but as good as anything to get back into the meat of things:

reworking the convoluted/buggy ASCII85 implementation...

also includes:

• some new test files
• filter output in the AST for otherwise uninterpreted streams
• elimination of a duplicated function (act_bytestream)
• removal of KUDU_OUTDIR from Makefile => that stuff goes to /tmp by default

TODO:

• validation of the "f <= UINT32_MAX" conditions
• unify act_a85group with act_a85partialgroup
• more test files (negative cases, especially)?
• are there any "impossible" encodings that we don't validate against, yet?!
• ...?