This allows to have the data generated with the correct user id
if provided in the configuration and doesn't match the current
user running the container.

This will allow to reuse the code in beat/beat.core.

Currently input and output cache files were "shared"
because of how the configuration was made.

Also, the input for the loop algorithm, although the same
as the loop_user, was not properly generated as a separate
input.

This reduces code duplication and will also
help clean up beat/beat.core> a bit.

This allows files with similar name not to clash.

For example input and input_data.

If process fails, return early and send done message to
other parties. This allows to stop everything more
cleanly.

Synchronous operation means that the following loop user
write will happen after the loop write method was called.

The end data index parameter allows to keep the output of
both block in sync.

This method will be called on the loop when
an output is written by the loop user algorithm.

This will be used to let the loop know that a
write occurred in the loop_user block.

This reflects more correctly what the test
is about.

Fix loop dataformat export

See merge request !56

This will ensure that exporting algorithm works correctly.

Currently the loops where skipped because of the way
optional group content where handled.

New loop user types

See merge request !55

This will give the developer the choice of how
they want their loop user to work.

Improve language auto discovery

Closes beat.editor#232

See merge request !53

Currently the auto discovery of the language doesn't take
into account usual errors related to loading a json file.
This patch adds more failure handling that should cover
these.

[doc] fixed external and internal links

See merge request !52

Handle duplicate key

Closes #25

See merge request !51