core: split block storage into workable units

[karalabe]

Superseded by #1779, though contains a pending commit so don't close for now

Background reading for this PR: database schema update for the fast sync algorithm.

This PR features the following:

• Header data access through accessor functions (allows caching and invalidating computed data).
• Split the block storage into separate headers and bodies (need pure header mode of operation).
• Support retrieving the raw RLP header/body data (during query, no need to parse and re-flatten).
• Refactor eth to use direct header/body accesses instead of entire block reads from the database.

WIP

[robotally]

Vote	Count	Reviewers
👍	0
👎	0

Updated: Tue Sep 15 14:37:26 UTC 2015

[obscuren]

Please rebase this PR on develop and keep relevant commits.

[karalabe]

I'll close this PR down for now. It contains one potential commit to make headers immutable, but let's try to work without that change and just make a mental note if we need it in the future.