[Tests] Add prioritisetransaction RPC test

[sdaftuar]

@morcos mentioned in #6898 (comment) that there's no rpc test for prioritisetransaction. This adds a simple test to exercise this logic.

In writing this test I discovered something unusual -- if you try to set a fee delta using prioritisetransaction that is negative and greater in magnitude than the transaction's fee, you get incorrect results. The CFeeRate constructor that takes a fee and size doesn't work if you pass in a negative fee. I don't really understand the details of how conversion and type promotion work in this kind of situation, but as far as I can tell, on my machine, I think the compiler promotes the int64_t nFeePaid to a uint64_t (nSize is a size_t):

nSatoshisPerK = nFeePaid*1000/nSize;

Link to the code:

bitcoin/src/amount.cpp

Line 15 in c983d6f

nSatoshisPerK = nFeePaid*1000/nSize;

I think this arithmetic may work on a 32-bit platform, because it looked to me like doing arithmetic with an unsigned int rather than a size_t appeared to work.

Anyway I expect that #6898 will fix prioritisetransaction to work properly with fee deltas that bring the modified fee below 0, so I just avoided that issue for now in this test; I can update it in the future to include that case after #6898.

[maflcko]

Nit: Shouldn't the print appear in front of assert? So when it fails, we know where it did (Not just by line number)?

[sdaftuar]

Well then it would be printing something incorrect if the assertion failed. I guess I could add a different print statement above the asserts, but not sure it's worth it? If the assertion fails I think the stack trace is informative enough, and I don't think the style here is inconsistent with the rest of the tests.

[maflcko]

Where would you put a comment, when you had put one? Then just replace the # with print and you get in L109:

print "Assert that prioritised transaction was mined"

[maflcko]

rest of the tests

I see no reason to do this. You could as well just remove the print completely.

[sipa]

utACK

[sdaftuar]

Rebased, moved prints

[laanwj]

Merged via 6abf6eb (needed trivial rebase change in rpc-tests.py)

[maflcko]

utACK 6abf6eb

[maflcko]

PR can be closed!?