This should reduce the occurences of travis failures on MacOS

Also fix some linter warnings

Debug traces to investigate a travis issue on MacOS

So that we only see the log message when we actually have to migrate.

downloader: different sync strategy

* trie: return hasher to pool

* trie: minor code formatting fix

* swarm: completed 1st phase of swap accounting

* swarm: swap accounting for swarm with p2p accounting

* swarm/swap: addressed PR comments

* swarm/swap: ignore ErrNotFound on stateStore.Get()

* swarm/swap: GetPeerBalance test; add TODO for chequebook API check

* swarm/network/stream: fix NewRegistry calls with new arguments

* swarm/swap: address @justelad's PR comments

* consensus/clique: fix comment typo

* eth,eth/downloader: fix comment typo

cmd, core, eth, light, trie: add trie read caching layer

Changed "signTransactions" to "signTransaction"

Changed "send" to "send," in two places

whishes --> wishes

cmd/swarm, swarm/api/http, swarm/bmt, swarm/fuse, swarm/network/stream, swarm/storage, swarm/storage/encryption, swarm/testutil: use pseudo-random instead of crypto-random for test files content generation (#18083)

- Replace "crypto/rand" to "math/rand" for files content generation
- Remove swarm/network_test.go.Shuffle and swarm/btm/btm_test.go.Shuffle - because go1.9 support dropped (see https://github.com/ethereum/go-ethereum/pull/17807 and comments to swarm/network_test.go.Shuffle)

params: update CHTs

On file access LDBStore's tryAccessIdx() function created a faulty
GC Index Data entry, because not indexing the ikey correctly.
That caused the chunk addresses/hashes to start with '00' and the last
two digits were dropped. => Incorrect chunk address.

Besides the fix, the commit also contains a schema change which will
run the CleanGCIndex() function to clean the GC index from erroneous
entries.

Note: CleanGCIndex() rebuilds the index from scratch which can take
a really-really long time with a huge DB (possibly an hour).

replace tabs by spaces in the code owners file

Access count was not incremented when chunk was retrieved
from cache. So the garbage collector might have deleted the most
frequently accessed chunk from disk.

Co-authored-by: Ferenc Szabo <ferenc.szabo@ethereum.org>

* p2p/simulations: Add custom services to simnodes + remove sim down conn objs

* p2p/simulation, swarm/network: Add selective services to discovery sim

* p2p/simulations, swarm/network: Remove useless comments

* p2p/simulations, swarm/network: Clean up mess from rebase

* p2p/simulation: Add sleep to prevent connect flakiness in http test

* p2p/simulations: added concurrent goroutines to prevent sleeps on simulation connect/disconnect

* p2p/simulations, swarm/network/simulations: address pr comments

* reinstated dummy service

* fixed http snapshot test

* rawdb: remove unused parameter for WritePreimages func and modify a
spelling mistake

* rawdb: update the doc for function WritePreimages

* metrics, p2p: add ephemeral registry

* metrics: fix linter issue

The lock in StateDB is useless. It's only held in Copy, but Copy is safe
for concurrent use because all it does is read.