transform() for mixed dataset

Nobody has claimed this yet.

Thanks lmcinnes for this great library!

I'm currently using umap on mixed data (and I've successfully used the code from issue #58 to implement mixed data fitting) and I was wondering how I could extend that to transform additional mixed data onto the fitted projection. Would the following implementation work?

Given that `umap_model_numerical` and `umap_model_categorical` are umap models respectively fitted onto the numerical and categorical (one encoded) data and that `df_numerical` and `df_categorical` are additional data to be projected onto the umap representation:

`fit1 = umap_model_numerical.transform(df_numerical.values)`
`fit2 = umap_model_categorical.transform(df_categorical.values)`
`intersection = umap.umap_.general_simplicial_set_intersection(fit1.graph_, fit2.graph_, weight=0.5)`
`intersection = umap.umap_.reset_local_connectivity(intersection)`
`embedding = umap.umap_.simplicial_set_embedding(fit1._raw_data, intersection, fit1.n_components,
fit1.learning_rate, fit1._a, fit1._b,
fit1.repulsion_strength, fit1.negative_sample_rate,
200, 'random', np.random, fit1.metric,
fit1._metric_kwds, False)`

From what I've seemed to understand, setting `'random'` for the `init` parameter for `simplicial_set_intersection` allows to not consider `fit1._raw_data` and `fit1._metric_kwds` parameters

Thanks in advance for your help and amazing work!

Start by reviewing UMAP's transform path and the mentioned entry points general_simplicial_set_intersection, reset_local_connectivity, and simplicial_set_embedding. Determine whether the proposed numerical and categorical workflow can support transforming additional mixed data, and define completion as a validated approach or a clearly scoped implementation with mixed-data examples.

Written by the indexing model from the issue text.