Amptk

* bug fixes for `amptk database`. Rewrote the dereplication function and added `--lca` or last common ancestor function for building the UTAX databases.
* bug fixes for `amptk SRA-submit` and update to edlib for searching.
* bug fix for `amptk filter` where samples passed to `--drop` are not used for index-bleed calculation
* bug fix for pre-processing and `--mult_samples` argument

* bug fix for `-t, --threshold` option of `amptk select` and `amptk remove`. Bug was introduced during last update when support for compressed files was introduced
* AMPtk now supports degenerate nucleotide primer matching, thanks to the very fast edlib v1.2 library. You will need to have at least edlib v1.2.0, can upgrade with pip, i.e. `pip install -U edlib`. The scripts will check your edlib version during runtime and let you know if you need to upgrade.

* bug fix for `amptk filter` where some mock sequences were not being annotated correctly
* upgrade `amptk SRA-submit` to use edlib alignment
* fix menu in several places

* Major update is to use edlib library for alignment, this is a dramatic increase in speed, however downside is that degenerate nucleotides are not supported in edlib currently (hoping to get this fixed soon). You can increase `--primer_mismatch` to allow for degenerate matches, keep in mind that currently any degenerate nucleotide will be counted as a mismatch. v0.9.3 still supports degenerate nucleotides, although the alignment is much less accurate and is 10X slower.
* edlib alignment now supports barcode_mismatches as well without a loss in speed.
* update to MergePE function, which allows user to select either vsearch or search for merging paired end fastq files, controlled via `--merge_method`. Update to phiX filtering to split files if >3GB to avoid memory problem in USEARCH 32 bit.
* add `amptk illumina3` method for pre-processing, this will demultiplex Illumina PE files along with index read files
* support for gzipped input files, as well as now default will output fq.gz demuxed files. Save space during processing.
* updated docker container and install instructions

* remove bedtools as dependency for converting BAM -> FASTQ. Now AMPtk will first try to use samtools if it exists, then bedtools if it exists, and will default to pybam native python parser to convert. Pybam is 10X slower than samtools, but is written in python thus no extra dependencies needed.
* added threshold filtering to `amptk remove` and `amptk select`, so you could remove all samples with reads less than 5000 by running, `amptk remove -i input.demux.fq -t 5000 -o output.demux.fq`

* bug fix for `amptk dada2` denoising where reads were getting ignored if they contained any ambiguous nucleotides. The filter for ambiguous nucleotides is still maintained prior to DADA2, note that terminal N's from padding will be properly removed, only internal ambiguous nucleotides are not allowed in DADA2