Using base64 as compression method resulting in double base64 encoding

[fangq]

This issue was reported by @kdreher in this mailing list post

https://groups.google.com/g/mcx-users/c/bToGluYYdao

it appears that Kris set -Z 2 which does not perform compression but only base64 encoding. the resulting .jdat file contains double-base64 encoded strings, making jdata decoding functions in Python and matlab fail (even though base64 only encoding was supported by these functions).

when a user asks to only apply base64 encoding, json-based output should skip a secondary base64 encoding.

an example to reproduce this bug is to run

mcx --bench cube60 -w xv -X 1 -F jnii -Z 2

and try to parse the output file

import jdata as jd
detp=jd.load('cube60_detp.jdat');