provide uncle size where available in RPC

[rphmeier]

cc @arkpar

Also adds a test for the Block and RichBlock types that size: None goes to null as expected.

[tomusdrw]

I think it should actually be 0, if we want to be aligned with geth.

[gavofyork]

geth is arbitrary and, in any sensible reading, flat-out wrong. null is the most correct answer here and we'll stick with it.

[tomusdrw]

@gavofyork @rob do I guess the change here should be reverted since it will return non-null values if the full.block is available for an uncle.

[rphmeier]

@tomusdrw Not sure if I follow, that's the whole point of the PR. The RPC isn't that well designed, so it is supposed to return the same result as eth_getBlockBy*, which includes a size. The size isn't always available, but we should return it when it is.

[tomusdrw]

So I guess it should consistently return null for uncles. Don't see a point in returning non-reliable size in any case.