[MSVermet]

Closes #477

Limit the categories in output multinouli distribution to the 10(?) categories with the largest probabilities.

Output would be like this (for this output I temporarily changed the limit to 2):

Column 9: "Embarked"
- Variable Type: categorical
- Data Type: Categorical
- Proportion of Missing Values: 0.0022
- Distribution:
	- Type: core.multinoulli
	- Parameters:
		- labels: ['S', 'C', ...]
		- probs: [0.7244, 0.1890, ...]
		- Showing 2 of 3 categories

Another option would be to change the output to something like this:

- Distribution:
    - Type: core.multinoulli
    - Categories: 25
    - Top 10:
        - 'A': 0,2410
        - 'B': 0,1830
        - 'C': 0,1200
        - 'D': 0,0850
        - 'E': 0,0610
        - 'F': 0,0300
        - 'G': 0,0250
        - 'H': 0,0200
        - 'I': 0,0150
        - 'J': 0,0100
    - ... 15 more categories (0,2100 combined probability)

[qubixes]

Thanks! It looks much better now. I think your proposal here is better than the alternative you mention. Otherwise it is going to take too much vertical space. Perhaps 10 is a little on the high side even, perhaps 7?

[qubixes]

I have a few more questions, but otherwise it's good. Feel free to merge after these are addressed.

[qubixes]

Why the repr(label) instead of just label?

[qubixes]

I guess because it's a np._str?

[MSVermet]

To make sure if label is not a string it can be joined. Not sure if this could ever happen, but just in case.