Something fails even without changes, due likely to emscripten docker
or travis system changes. This is too undeterministic to rely upon so
we're moving the testing infrastructure over our own Jenkins based CI
for DECODE.

function declaration wasn't consistent, triggered warming in
emscripten build pointing to unpredictable behavior. also now made
loading of libraries less verbose.

correctly loads files when configured (typo s/env/enc)
accepts data as second argument

now prints to stderr when executed by read_json
added new error() function to print out to stderr from lua
added write_json() to wrap inspect.print() for DX

-tag library name with ARCH VERSION BRANCH and HASH to get pedantic version tracking

also avoided some code replication in lua_modules with the
zen_exec_extension function.

adds commandline '-p' option to parse source into an Abstract Source Tree and
then return it in JSON format to the output. Uses lua-parser and lpeglabels
which are now embedded.

now when printing ecp a json is produced with the curve point and
metadata about curve type and software versions producing it. This is
useful for debugging purposes and to track regressions.

also adds some minor umm memory check

now keyring:private(octet) method generates the public key upon
setting the private inside id.

still receiving an error if this free is operated in js, so will build
an exception around it for now.

this is an interesting case where a system free() call on a pointer
already freed by lua as lightuserdata on lua_close() is not leading to
a crash on any build but js builds (more sensitive to this).

caused indeed by saving the Z context inside L and freeing after
freeing L. Now should be OK, but good to keep an eye on teardown after
usage of deeped init/exec/teardown API takes place.