Add RPC commands and tests for block reorganizations

[dexX7]

This PR adds the required RPC calls (and some related) to test block reorganzations.

The tested transaction types and aspects of Omni Core are:

• simple send
• send to owners
• property creation

There are two significant downsides: the "invalidateblock" command was introduced with Bitcoin Core 0.10, and is therefore not available on the CI server, but worse: I had no handy way to clear the memory pool, and added an extra extension for this.

All reorganization tests are skipped, if "clearmempool" is not available, or if "setgenerate" doesn't return a list of block hashes, indicating a client version, which is not based on 0.10.

I'm not sure, if this PR should be merged, given the context. Ideally, there would be a workaround to clear the memory pool, the client version properly detected, and tests ignored otherwise, instead of being skipped.

[msgilligan]

Maybe I should have read this more carefully before merging, but worse case I can @ignore the tests.

[msgilligan]

Skipping the tests works just fine. I want to migrate the CI server to 0.10.0 soon, so that gets us part of the way there.

What's the likelihood of the clearmempool extension making it into either Omni Core or upstream (Bitcoin Core)?

I marked BaseReorgSpec as abstract: b00cdff This will prevent the test runner from running that class as a test.

@dexX7 Can you submit an additional PR with some of your notes from above in the JavaDoc comments for the new methods in BitcoinClient.java? (e.g. Which calls require 0.10.0 and which ones require your "extra extension")

[dexX7]

Reorganziation tests are important, and there are likely going to be more in the future, but I'd consider adding "clearmempool" as last resort, and I have some ideas for alternatives, which we might do in OmniJ directly.

@zathras-crypto: maybe you have an idea, how to "force" clearing the mempool? The idea I had was to explicitly "taint" transactions with coins from a block, which is later going to be invalidated, and to trigger mempool rejections as result, because the original source of coins would no longer be valid.

I'm going to push comment improvements tomorrow, and while we're at it: @msgilligan: Omni Core 0.0.10 has a bunch of new commands, such as "trade_MP", "getorderbook_MP", ... , and currently pending (actually waiting for more testing), also transaction creation commands for the missing transaction types. I didn't add them yet, because it's not clear to me, how different client versions are to be handled, but starting with clear comments is a good idea.

[msgilligan]

Given that OmniJ is "pre-alpha" and the API can change "without notice", that Bitcoin 0.10.x has been available for some time now, and that the Omni Core testing "use case" of OmniJ will be focusing on 0.10.x very soon, it seems the general strategy should be to move to 0.10 APIs as quickly as possible.

Since we are still using 0.9.x on the CI server and Omni/Master Core 0.9.x is still in production, we should maintain backwards compatibility with 0.9.x in the RPC clients until 0.10.x ships.

This means we can add the new commands at any time as long as they're documented as requiring 0.10.x.