pylint 1.9.3 -> 2.0.0 jump results in some linter warnings

A try-except-raise checker was added to pylint in 2.0.0 (originally in this commit), raising a warning (error downgraded to warning in 2.1) if there's an immediate re-raise in an except, outside of some special cases.

Here's one of the offending constructions (or see it in context here)

        try:
          # ....
        # Unknown role, re-raise exception.
        except tuf.exceptions.UnknownRoleError:
          raise
        ## Note: no further except or else clauses!

It's not clear to me why this construction is useful, beyond the guess that it might indicate to a reader that an exception of a certain type is expected to occur here in certain situations.... That alone doesn't seem worth the extra bit of code and the wrinkle in the exception stack, though.

Is anyone aware of a good reason to keep a construction like the above?
@alyptik @SantiagoTorres
(Note that if there are further except clauses, there are reasons to have an immediate re-raise -- see pylint 2323 -- the case of excluding a subclass re-raise from a broader exception handling clause. Incidentally, this would still be something to avoid, but it does make some sense....)

PR #771 fixes the code construction issues pylint > 2.0.0 finds. Since this PR does not introduce any of those issues, it doesn't need to wait to be merged until after 771.

As best I can tell, updating these dependencies doesn't break anything, so I could use a review at this point and I'll add sign-off and merge without the linter passing (see above comments).

@trishankatdatadog, can you tell me if any of these dependency changes would be an issue for you?

Updating cryptography shouldn't bother us. Thanks for checking!