Add back processed transfer count to service statistics

[zdave-parity]

This was removed when accumulate was combined with on_transfer. It seems like useful statistic, so add it back.

[davxy]

In 13.7, $a$ domain need to be fixed:

Currently:

But according to S (12.29), which is used in the assignment 13.12, it needs to be $a \in (\mathbb{N}_G, \mathbb{N}, \mathbb{N})$

If we change the gas to come first, then for the sake of consistency, it would be nice to change it in $r$ as well: $r \in (\mathbb{N}_G, \mathbb{N}))$

[zdave-parity]

Thanks, good spot. I don't think I changed the order of the stats; it seems like the type definition has always been the opposite way round to the tuples in \accumulationstatistics. I see that in PolkaJam the stats struct currently has gas last and presumably other implementations match that. Additionally the refine stats have gas last. So I'll flip the accumulation tuple around to match.

[zdave-parity]

It's somewhat disappointing that no one has pointed this error out. It suggests that people are simply updating their implementations to match PolkaJam rather than the GP.

[jaymansfield]

It suggests that people are simply updating their implementations to match PolkaJam rather than the GP.

Not really at all. If an implementation encoded C(13) according to the πS definition in 13.7 (with gas last), they are following what is defined in the GP according to that specific formula (regardless if its correct or not), not PolkaJam.