Fix ReadPair's never being freed

[Donwulff]

Fixes memory leak in PairEndProcessor::interleavedReaderTask() where ReadPair objects were not being freed, resulting in a leak of 16 bytes per read pair across the input (Issue #392 ).

Changes:

• Modified call signatures to reuse the same ReadPair for all reads, as it's a return container for the Read pairs.
• Added a method to detect the end of input for clarity.
• Updated freeing of the Read arrays to utilize array delete.

Performance Metrics Comparison on 1KG HG00312 exome with 110M reads:

Metric	Before	After
User time (seconds)	1524.60	1507.94
Maximum resident set size (kbytes)	4423400	2548668

All results remained identical.

[bwlang]

I observed memory leaks as well when running fastp like this:

fastp --stdin -l 2 -Q --trim_poly_g --thread 1  -j library1.fastp.json --split_by_lines 8000000 --interleaved_in --out1 library1.1.trimmed.fastq --out2 library1.2.trimmed.fastq

after applying this commit and rebuilding

make LIBRARY_DIRS=$CONDA_PREFIX/lib INCLUDE_DIRS=$CONDA_PREFIX/include -j 32
patchelf --set-rpath $CONDA_PREFIX/lib ./fastp
cp fastp $CONDA_PREFIX/bin

the problems we were seeing with fastp getting killed by the cluster manger(exceeded 16G) have been resolved.

@sfchen : any concerns about merging this PR?