Improve typing in phirgen, fix measure args bug, pretty print PHIR

[qartik]

• Improve typing in phirgen (allow Any)
• Fix package inclusion of missing files
• Log prettified PHIR, fix bug with measure args

Sample outputs:

# pytest tests/test_api.py::TestApi::test_pytket_to_phir_h1_1
PHIRModel(
    format_='PHIR/JSON',
    version='0.1.0',
    metadata={'source': 'pytket-phir'},
    ops=[
        QVarDefine(metadata=None, data='qvar_define', data_type='qubits', variable='q', size=2),
        CVarDefine(metadata=None, data='cvar_define', data_type='i64', variable='c', size=2),
        Comment(c='Rz(1) q[0];'),
        QOp(metadata={'angle_multiplier': 'π'}, qop='Rz', returns=None, args=[['q', 0]], angles=[1.0]),
        Comment(c='PhasedX(0.5, 0.5) q[0];'),
        QOp(metadata={'angle_multiplier': 'π'}, qop='PhasedX', returns=None, args=[['q', 0]], angles=[0.5, 0.5]),
        Comment(c='Rz(1) q[0];'),
        QOp(metadata={'angle_multiplier': 'π'}, qop='Rz', returns=None, args=[['q', 0]], angles=[1.0]),
        Comment(c='Rz(1) q[1];'),
        QOp(metadata={'angle_multiplier': 'π'}, qop='Rz', returns=None, args=[['q', 1]], angles=[1.0]),
        Comment(c='PhasedX(0.5, 0.5) q[1];'),
        QOp(metadata={'angle_multiplier': 'π'}, qop='PhasedX', returns=None, args=[['q', 1]], angles=[0.5, 0.5]),
        Comment(c='Rz(1) q[1];'),
        QOp(metadata={'angle_multiplier': 'π'}, qop='Rz', returns=None, args=[['q', 1]], angles=[1.0]),
        Comment(c='PhasedX(1.5, 0.5) q[1];'),
        QOp(metadata={'angle_multiplier': 'π'}, qop='PhasedX', returns=None, args=[['q', 1]], angles=[1.5, 0.5]),
        Comment(c='ZZPhase(0.5) q[0], q[1];'),
        QOp(metadata={'angle_multiplier': 'π'}, qop='ZZPhase', returns=None, args=[['q', 0], ['q', 1]], angles=[0.5]),
        MOp(metadata={'duration': 0.0}, mop='Transport'),
        Comment(c='Rz(1.5) q[0];'),
        QOp(metadata={'angle_multiplier': 'π'}, qop='Rz', returns=None, args=[['q', 0]], angles=[1.5]),
        Comment(c='Measure q[0] --> c[0];'),
        QOp(metadata=None, qop='Measure', returns=[['c', 0]], args=[['q', 0]], angles=None),
        Comment(c='Rz(0.5) q[1];'),
        QOp(metadata={'angle_multiplier': 'π'}, qop='Rz', returns=None, args=[['q', 1]], angles=[0.5]),
        Comment(c='PhasedX(1.5, 0.5) q[1];'),
        QOp(metadata={'angle_multiplier': 'π'}, qop='PhasedX', returns=None, args=[['q', 1]], angles=[1.5, 0.5]),
        Comment(c='Measure q[1] --> c[1];'),
        QOp(metadata=None, qop='Measure', returns=[['c', 1]], args=[['q', 1]], angles=None),
        MOp(metadata={'duration': 0.0}, mop='Transport')
    ]
)

# python tests/e2e_test.py
PHIRModel(
    format_='PHIR/JSON',
    version='0.1.0',
    metadata={'source': 'pytket-phir'},
    ops=[
        QVarDefine(metadata=None, data='qvar_define', data_type='qubits', variable='q', size=3),
        CVarDefine(metadata=None, data='cvar_define', data_type='i64', variable='c', size=3),
        Comment(c='CX q[0], q[2];'),
        QOp(metadata=None, qop='CX', returns=None, args=[['q', 0], ['q', 2]], angles=None),
        Comment(c='H q[1];'),
        QOp(metadata=None, qop='H', returns=None, args=[['q', 1]], angles=None),
        Comment(c='Measure q[1] --> c[1];'),
        QOp(metadata=None, qop='Measure', returns=[['c', 1]], args=[['q', 1]], angles=None),
        MOp(metadata={'duration': 2e-06}, mop='Transport'),
        Comment(c='H q[0];'),
        QOp(metadata=None, qop='H', returns=None, args=[['q', 0]], angles=None),
        Comment(c='Measure q[0] --> c[0];'),
        QOp(metadata=None, qop='Measure', returns=[['c', 0]], args=[['q', 0]], angles=None),
        Comment(c='H q[2];'),
        QOp(metadata=None, qop='H', returns=None, args=[['q', 2]], angles=None),
        Comment(c='Measure q[2] --> c[2];'),
        QOp(metadata=None, qop='Measure', returns=[['c', 2]], args=[['q', 2]], angles=None),
        MOp(metadata={'duration': 0.0}, mop='Transport')
    ]
)

[qartik]

Previously, both the qbit and cbit were part of the args for Measure.

[peter-campora]

No better types we can give than Any?

[qartik]

With the current approach, i.e., constructing Python dict's instead of a PHIRModel object, the best we can do is Any. Someday, we can consider directly constructing PHIRModel but I found it difficult to go that route in my first attempt.

[peter-campora]

Looks good. My only comment was about whether we could replace Any with a different annotation, but we can make changes like that gradually.