Remove mostly unused util.py

[lukpueh]

• addresses part of #731
• blocks on #772 (commits are included in this PR)

closes #245

---

The util.py module provides a mix of minimal functions related to local files and directories. These functions are mostly unused, exceptions are (according to sourcegraph.com):

• persist_temp_file in python-tuf (cc @jku)
• get_file_hashes in scai-demos (cc @marcelamelara, @adityasaky)
• get_file_details, load_json_file in taf (cc @renatav, @adityasaky)

It seems feasible to reimplement the desired functionality downstream.

The benefits of removing util.py are:

• minimizes the API surface
• removes one dependent of the about to be removed schema.py (#183)
• makes unittest_toolbox obsolete

[adityasaky]

scai-demos can maybe use the API provided in in-toto's runlib to calculate file hashes? cc @marcelamelara

I'm not involved in taf atm but tagging @patzielinski in addition to Renata.

[lukpueh]

scai-demos can maybe use the API provided in in-toto's runlib to calculate file hashes? cc @marcelamelara

Happy to help!

I'm not involved in taf atm but tagging @patzielinski in addition to Renata.

taf at least shouldn't be disrupted, as they pin an older version of securesystemslib.

[renatav]

In TAF, load_json_file is used in one (very old) test and we can easily use our own implementation of that functionality instead. I see no problems with implementing get_file_details in TAF too. We still have version pinned to 0.25.*". Thanks for the heads up!

[marcelamelara]

scai-demos can maybe use the API provided in in-toto's runlib to calculate file hashes? cc @marcelamelara

Took a look at this, and I think that the digest_filename function from securesystemslib.hash may be a better fit. I'll have to convert the hash object into a dict, but that should be relatively trivial.