Closes #2855: Read batches for Parquet strings

[bmcdonald3]

In the first implementation of Parquet, strings were the only data type being read one at a time, rather than in batches since we can't read directly into a uint(8) buffer, so we were reading one value at a time and then copying it into the Chapel buffer. Rather than doing that one a time, we can switch to reading batches and then bulk copying to improve performance. Here are some performance numbers collected on 2 nodes of a Cray XC with varying batch sizes:

batchSize	exec time
1	9.63s
10	3.66s
256	3.05s
512	3.03s
8192	3.05s

So, we see about a 3x improvement.

Closes #2855

[bmcdonald3]

I believe the test that is failing is because of the writeMultiCol parquet function was not accounting for null values and thus didn't care about the definition level, which we now have to use to detect null values, since we can no longer just check values_read to give us indices of where there may be a null string.

I think it will require some modifications to the writeMultiCol function in order for this to work.

[stress-tess]

Looks good to me!! Thanks @bmcdonald3!