Fix deadlock for garble-vm

[th4s]

This PR fixes the VM hanging, when circuit output ranges are used as input ranges for a new circuit. The change fixes the curently hanging tests of the prf and cipher crates of tlsnotary.

When a circuit is executed the input range corresponding to the output range of the circuit output is not marked as complete. This leads to is_committed returning false and the circuit never being executed.

Questions

1. Do we also need to touch assigned or all of InputView? Anything else which needs to be adapted for consistency?
2. Is this change compatible with the overall picture, because I get the idea that so far inputs and outputs are treated entirely different.

[sinui0]

This isn't the right solution, as we do need to distinguish between input and output values. The right approach is likely to modify is_committed so that it considers completed outputs as committed.

[th4s]

Adapted is_committed now instead of mixing input and output ranges. I also included Dan's suggestion to fix the preprocessing in this PR.