Make parser with max block height more efficient

[hkalodner]

BlockSci shouldn't need to run the 1st parser stage over the entire blockchain if the max block is substantially lower than the number of blocks in the node. Instead it should cut off the block header scan when it is confident it has found the longest chain up to the requested length.

Expected Result

The first stage parser should only have to scan up to a reasonable confidence threshold to assume it is looking at the longest chain.

Actual Result

The first stage parser scans every existing block.

Reproduction Steps

Not Applicable

System Information

Not Applicable

[hkalodner]

This is in response to #158 pointing out this problem