Testlib: remove the “partial” feature if speed is no longer an issue

[castulo]

Many of the testlib functions support a -p flag which stands for "partial". The main purpose of this flag is to skip running some of the steps in a function more than necessary.
For example after updating a manifest we need to remove the outdated manifest tar and we need to generate a new one based on the updated manifest. We also need to update the hash of the updated manifest into the MoM, recreate the MoM's tar and sign it. If we need to make many updates to the same manifest it doesn't make sense to go through all these changes every time an update is made, only after the last update has been done we would finally recreate the tar.

Currently this is what we would do:

update_manifest -p $manifest some_change
update_manifest -p $manifest another_change
update_manifest $manifest last_change

That was a necessary option to try to speed up tests, however is error prone because we could forget to run the last update without using the -p flag, which would cause corruption in the test environment.

At the moment we are evaluating other options to speed up tests (for example caching the test environment and its test resources), once that work is done we can probably get rid of the -p "partial" flag in all functions to remove that point of error, since the time necessary to create an environment won't be an issue any more.

[otaviobp]

Speed is still a concern and it would be way better if we didn't need to recalculate hashes multiple times.
The current API for update_bundle is prone to errors because if you use -p for every command, it won't work and if you don't use -p at all, it will not be optimized.