fix: correct repr for MultiCell

[henryiii]

Followup to #1008. This fixes the repr for MultiCell, which currently always shows 0, since we can't get the actual instance, but just it's type (without the value of nelem).

I've added a h.get_storage() method which gets a storage object with the same parameters, an instance of h.storage_type. It's natural for the repr, and allows users to query with h.get_storage().nelem. Thoughts, @HDembinski and @Superharz? I could make it a property instead, such as h.storage.nelem. The main issue is it is just the bh.storage creation object, and doesn't actually give access to storage, so you can't do much with h.storage.

Edit: I realized that making this a property actually works better for pattern matching, which expects properties to match constructor arguments, so I've used h.storage now.

I thought about exposing .nelem on the histogram, but I can't properly statically type it (mypy doesn't allow @property and @typing.overload to be combined), and it's just None on most histograms.

[Superharz]

Hi @henryiii, I am definitely in favor of having the correct number of cells contained in the representation. I also noticed that right now it always says MultiCell(0) regardless of the specific number of cells.

[HDembinski]

Lgtm