Support in-place numerical operations for SparseLabelOps

[mrossinek]

What is the expected enhancement?

#860 suggest supporting in-place operations on SparseLabelOp instances. That issue focuses on the addition of an inplace keyword argument to the public API where this can be useful.

This issue suggests to take this one step further and implement similar methods to allow in-place numerical operations such as addition, scalar multiplication, etc.This can be done by implementing the dunder methods like __iadd__ which get called when doing a += b (currently this gets expanded to a = a + b).

[kevinsung]

Some observations:

• The type of SparseLabelOp will change from Mapping to MutableMapping (or even dict)
• Setting copy=False in SparseLabelOp.__init__ won't be supported. Because the argument data can be a Mapping which may be immutable, we will need to copy the data no matter what. So we should just remove the copy argument.
• Alternatively, we can keep the copy argument but change the type of data from Mapping to MutableMapping (or dict)