Bugfix for drawing RandomEncodingCircuit

[kit0kat]

There seems to be a bug when drawing a RandomEncodingCircuit.

Minimal re-produceable code (run in a jupyter notebook with matplotlib installed):

1. Passing no num_features:

from squlearn.encoding_circuit import RandomEncodingCircuit
RandomEncodingCircuit(num_qubits=4, seed=42).draw("mpl")

❌ TypeError: '>' not supported between instances of 'NoneType' and 'int'

2. Passing num_features only to Circuit():

from squlearn.encoding_circuit import RandomEncodingCircuit
RandomEncodingCircuit(num_qubits=4, seed=42, num_features=3).draw("mpl")

❌ TypeError: '>' not supported between instances of 'NoneType' and 'int'

3. Passing num_features only to draw():

from squlearn.encoding_circuit import RandomEncodingCircuit
RandomEncodingCircuit(num_qubits=4, seed=42).draw("mpl", num_features=3)

⚠️ no error, but drawing shows x instead of x[0...3] vector

4. Passing num_features to Circuit() and draw():

from squlearn.encoding_circuit import RandomEncodingCircuit
RandomEncodingCircuit(num_qubits=4, seed=42, num_features=3).draw("mpl", num_features=3)

❌ ValueError: Number of features in the input data (1) does not match the number of features in the encoding circuit (3).

This PR proposes a possible fix for the draw() method that seems to work for all 4 cases.

[MoritzWillmann]

Hey @kit0kat, thanks for catching and fixing this. I'll get this merged ASAP 😄