Amptk

* Add phix filtering for Illumina data. As part of the PE merging function in `amptk illumina` and `amptk illumina2`, scripts will also now run phix removal.
* Workaround for DADA2 error where samples that only have 1 read post filtering trigger a `derep$quals matrix` error. `amptk dada2` now has `-m, --min_reads` option to drop samples that have fewer than `-m` reads. Default this is set to 10, however, in practice probably this should be much higher, but this should avoid the above error.

* added better support for `amptk SRA-submit`
* added ability to normalize heat map
* added `amptk SRA` which can be used to process reads downloaded from the SRA, where they are in a single FASTQ file, i.e. from ION or 454 data that has been demultiplexed into samples and then submitted.
* created Dockerfile for using `amptk` with the `scipy-notebook` jupyter notebook server.

- unify the output naming files from UNOISE2 and DADA2 "clustering" output.

- support for new DADA2 algorithm allowing variable length reads, must have > v1.3.3.

- add `amptk drop` to remove OTUs from a dataset and then create an updated OTU table
- fix for `amptk illumina` where empty files would cause script to terminate
- fix for biom output to explicitly be json
- fix in `amptk remove` to allow fasta output

- bug fixes for pre-processing steps where short primer-dimers could make it through filtering, get padded with N's and get incorporated as OTUs in clustering
- update to `amptk filter` to output the final OTU table to have real read counts as opposed to "pseudo" counts from normalization. Filtering is done with normalization, but now read counts are restored to original read numbers. Important for downstream stats like beta diversity
- improved read summary reporting in pre-processing steps
- update to `amptk unoise2` to output both inferred or denoised sequences/tables as well as biological OTU sequences (clustered at 97%).