chifra init: high number of pool items in chifra init causes repeated "UNKNOWN ERROR" against Pinata

[tjayrush]

We used to have nCPUs * 2 as the number of go routines in the pool that queries Pinata and we were getting quite a few INTERNAL or UNKNOWN error returns.

I turned that value down to nCPUs / 2 and the problem went away.

Not quite sure if this was Pinata or some code on our side. When we get a chance, it would be good to track this down and either handle rate limiting better (quadratic backoff) or help Pinata figure out what's happening.