Fixed bloom_filter_bitwise test (issue #12)

[edenzik]

In a bitwise bloom filter test, the test passed with an additional hash function at the first level.

[mavam]

Thanks for the patch. To me, it looks like the errors of #12 concern the stable Bloom filter, not the bitwise. Were you also getting errors for the stable Bloom filter?

But in any case, there appears to be some non-determinism in the test suite, presumably due to standard library and architecture (32 vs 64 bit) differences.

What I'd really like to move towards is a different architecture with the hashing all together, as I've started in the topic branch N3980. This form of hashing already works in another project of mine and would have just to be ported over in this project. But as I'm lacking the cycles to do so, any help from the community would be much appreciated. (I'm happy to provide feedback/help throughout the process.)