XEth cleanup

bugfix with idle too long triggered after peer switch

* In case of uncle mining (e.g. same TS) the miner would stop if all
  threads happened to mine a potential uncle

- reset bestIdleTimer when launching head section process
- log chain head events
- common.Hash key in activateChain map

Update ethash lib

* Refund of value

a7081bc update tests
d2678cb update random tests
ca0401b update tests

git-subtree-dir: tests/files
git-subtree-split: a7081bc54e9d72cc480218cb4b218dbb04c7cde7

* Value XFER are refunded back to the sender if the execution fails