predict method doesn't correctly form basis matrix

[bmeyers]

The predict method uses default setting when extending the basis matrix to new times. So, when using a "standing wave" dictionary, the method does not construct the correct extension.

Fix: have constructor read all kwargs from init method, e.g., standing_wave.

[bmeyers]

Trying to transform data outside the training window, when using a standing wave basis:

spq2.inverse_transform(new_data, new_ix + len(transform2))

Results in the following error:

Cell In[55], line 1
----> 1 plt.plot(spq2.inverse_transform(new_data, new_ix + len(transform2)))

File [~/Documents/github/spcqe/src/spcqe/quantiles.py:287](http://localhost:8889/lab/tree/covariance_matrices/~/Documents/github/spcqe/src/spcqe/quantiles.py#line=286), in SmoothPeriodicQuantiles.inverse_transform(self, X, y)
    285 # get correct basis matrix and quantile estimates for time period of prediction
    286 if y is not None:
--> 287     new_quantiles = self.predict(y)
    288 else:
    289     new_quantiles = self.fit_quantiles

File [~/Documents/github/spcqe/src/spcqe/quantiles.py:172](http://localhost:8889/lab/tree/covariance_matrices/~/Documents/github/spcqe/src/spcqe/quantiles.py#line=171), in SmoothPeriodicQuantiles.predict(self, X, y)
    170 newb = self.extend_basis(np.max(time_index) + 1)
    171 newb = newb[time_index]
--> 172 newq = newb @ self.fit_parameters
    173 newq = np.sort(newq, axis=1)
    174 return newq

ValueError: matmul: Input operand 1 has a mismatch in its core dimension 0, with gufunc signature (n?,k),(k,m?)->(n?,m?) (size 112 is different from 217)```

Method works correctly when new index is inside old index range.