Optimize Syncer

[patrick-ogrady]

Related Issue: coinbase/mesh-cli#87

This PR greatly improves sync performance by changing syncing to be continuous instead of discrete (syncing in 1000 block chunks). Most performance gains come from keeping our IO resources in high use (previously, we wouldn't attempt to store anything while fetching the block range). I will post a chart describing the magnitude of sync speedup when merging this into the rosetta-cli.

As a side effect, we greatly reduce memory usage by avoiding caching up to 1000 blocks in memory before processing (when the default concurrency is used, this is a memory usage reduction of 99%).

Changes

• Make syncer continuous instead of discrete
• Improve syncer test coverage
• Change BlockConcurrency to SyncConcurrency
• Remove BlockConcurrency from Fetcher
• Cleanup logic in syncer and in tests

[coveralls]

Pull Request Test Coverage Report for Build 3279

• 87 of 99 (87.88%) changed or added relevant lines in 2 files are covered.
• 1 unchanged line in 1 file lost coverage.
• Overall coverage increased (+4.3%) to 69.833%

---

Changes Missing Coverage	Covered Lines	Changed/Added Lines	%
syncer/configuration.go	4	8	50.0%
syncer/syncer.go	83	91	91.21%

Files with Coverage Reduction	New Missed Lines	%
syncer/syncer.go	1	86.49%

Totals
Change from base Build 3150:	4.3%
Covered Lines:	2169
Relevant Lines:	3106

---

💛 - Coveralls

[qiwu7]

👍