Add Steane-type and teleportation-based QEC for the Steane code

[perlinm]

Teleportation-based QEC is essentially just one-bit teleportation, in which only one type of error (X or Z) is preserved in the teleported state. Teleporting in two different ways thereby corrects all errors.

I used state vector simulations to verify that these circuits correct all physical single-qubit errors by:

1. Initializing a logical Steane qubit in a random logical state.
2. Applying a single-qubit X, Y, or Z error.
3. Running a qubit.steane_qec cycle.
4. Asserting that the (physical) state vector of the logical qubit is equal to the state vector before the applied error (up to global phase).

I am not entirely certain that I am using/storing classical registers correctly, so @ciaranra please take a look.

Finally: the methods here write a flag qubit to remember whether there were non-trivial syndromes. It may make sense to "guarantee" that this flag qubit is set correctly and remove this suggestive comment. Thoughts?

[perlinm]

Something funny is going on in my testing, so I'm going to mark this as a draft until I figure it out...

UPDATE: I think I fixed it. I'm still not sure I'm setting classical registers correctly though.

[perlinm]

I also renamed flag_bit --> flag in this PR, for consistency with conventions elsewhere in the repo.

[perlinm]

I think this PR fixes #112, which appears to just be an incorrect type hint

[ciaranra]

Good questions. At first glance everything looks good... but I'll comb over everything shortly.

More future facing stuff (not this PR):
I am putting the final touches on SLR->PHIR, which should allow for establishing proper internal tests for most of the code (might need to think about non-Clifford stuff). Hopefully that will help confirming behavior. Also, probably would be good to do a documentation sweep and check/establish for style consistency and provide guides in the documentation so we can keep straight how things are approached.

[ciaranra]

LGTM