Increase test coverage

[simleo]

This PR expands test_readfile and adds test_{createdp,pushtopandas}.

At present, since:

• the code is in a tightly coupled state and
• the exact specs are still a matter of discussion (Links vs Tracks and Linearity Tracks#18),

it's hard to build small artificial test cases where the expected behavior can be determined by visual inspection. On the other hand, refactoring / cleaning up the code without adequate test coverage can lead to regression. For this reason, the approach adopted in this PR is to use the current examples as a reference for the expected results, which basically amounts to assuming that the current code (which generated those examples) is correct. Note that I've added tracks.csv files since they were necessary for testing the pushtopandas module. They have been generated with a quick hack:

import os
import csv
from biotracks.pushtopandas import push_to_pandas

dirs = [
    "examples/CellProfiler/example_1/output/dp",
    "examples/CellProfiler/example_2/output/dp",
    "examples/ICY/example_1/dp",
    "examples/ICY/example_2/dp",
    "examples/TrackMate/example_1/dp",
    "examples/TrackMate/example_2/dp",
]

for d in dirs:
    if "TrackMate" in d:
        obj_id = 'SPOT_ID'
    elif "ICY" in d:
        obj_id = 'OBJECT_ID'
    else:
        obj_id = "ObjectID"
    dict_ = push_to_pandas(d, obj_id)
    dict_['tracks'].to_csv(os.path.join(d, 'tracks.csv'), index=False, quoting=csv.QUOTE_NONE)

[pcmasuzzo]

This PR is a substantial move forward towards bigger test coverage. +1 in particular for adding test on creation of the datapackage for the three software currently covered. All good to me! Thanks.

[simleo]

Coming up next: simleo/biotracks@168fb30. It depends on this PR, so I'll wait for this to be merged then rebase and open a new PR.