Cblaster

* Fall back to built-in `sqlite3` module when importing `genomicsqlite` fails

Thanks to brymerr921:

**Bug fixes:**
- By default, diamond blastp only saves a maximum of 25 hits per query, severely limiting the number of blast hits reported. Large databases are especially impacted. To solve this, the `--hitlist_size` is now passed on `diamond` inside `cblaster search -m local`. (1c16caa) (See issue 75)
- Added an option to `cblaster makedb` allowing users to provide a single `txt` file containing a list of input files. This solves the issue where the operating system limits the length of wildcard expansion on the command line. (bb9085c)
- Fixed error that arose when specifying `sum` or `max` for `-bkey` (fee315e) (see issue 76, thanks to chasemc)

**Enhancements:**
- `diamond makedb` now uses the number of CPUs requested by the user (instead of all available CPUs) (0db55cc)

**File compression:**
- Use genomcsqlite3 to reduce the size of the final sqlite3 database by 40-60% (bae6b31)
- `cblaster makedb` accepts gzipped input files (241894d)
- `cblaster makedb` now writes a fasta.gz output file (241894d) and informs users they will need to decompress first if running `hmmsearch` (4c63fa5)

* Fix `extract_clusters` module not working for HMM searches (https://github.com/gamcil/cblaster/commit/2dcae875ba8bf170c1bba7d55fe13305317a2859)

* Explicitly check for ``decode`` method on HTTPResponse objects (as well as status code 200) from IPG Efetch to prevent AttributeError (62)

* Change default value for ``--percent`` argument to 0 from 50 to prevent confusion when no results
* Fix parsed GFF records not being correctly converted to SQL insertion tuples, should address 63

* Adds the ability to load local profile HMM files instead of just Pfam domains
* Can now look up Pfams by family name, e.g. --query_profiles AMP-binding
* Add local profile HMM support to GUI
* Add example search output and small README on how to generate it
* Change default font family of generated plot