refactor: rio parsing

[cmdoret]

This PR refactors the whole triple parsing and pseudonymization to use oxigraph's rio types.

It does the following:

1. Add a Pseudonymizer trait to allow easy pseudonymization of subjects and objects in the future.
   It is implemented for MaskedTriple and the rio base types (Subject and Term).
   When calling pseudo() on a masked triple, it forwards the call to the masked components to call their respective pseudo() implementation (although they currently do nothing).
2. Replace the main parsing loop with rio's low level parse_step() method, which takes care of reading the next triple and should support multiple formats, including when 1 line != 1 triple.

[gabyx]

@cmdoret : Wuu, now we are getting fancy. But I think its good. I have a look at it.

[gabyx]

This is really nice I think.
Well done.

One thing: Read the comment with triple.pseudo(mask) first.
Might help, then the rest can be ignored.

[cmdoret]

Thanks for the thorough review, you made very good points and I did quite a bit of refactoring in crate::model based on it, learning a lot in the process 😃 .
The code is definitely better, but I was hesitant between two alternatives:

We now have a TriplePart enum which can be cast to/from u8 and a TripleMask. Alternatively, we could have only TripleMask and implement many small methods has_subject, set_subject, has_predicate, ... which I found a bit redundant (i.e. something along these lines).

Not sure if there is even a better solution. At least in terms of performance it should be quite good; the mask is just a u8 and accessor/setter methods perform a bitwise operation 🚀 .

[gabyx]

Looks good to me!
Way better now.

The bitflag thing of yours might be good too and actually nice when we could write a for loop in the pseudonimize loop. Without that there is not much benefit over using bitflags ! (rust macros are not evil: check codes ability to expand them rust-analyzer but jeah sometimes they make things less boilerplaty but also confusing.

You can choose.
Happy wit both!