Toulligqc

* Generated images were not included in the main HTML report file.

* Add some flexibility to barcode specification (Thanks to Hunter Cameron).
* The sequencing_summary.txt and sequencing_telemetry.js files can now be read compressed in gzip or bzip2.
* Add some log information on stdout for duration of the sequencing summary extractors.
* Change the format of the duration in log.
* Now logs memory used by dataframes.
* Fix Docker image build issue when updating setuptools.
* QScores and durations are now stored in 32 bits floats instead of 64 bits to reduce memory consumption (≈25% for 1D data).
* Barcode arrangements are now stored as categories instead of strings.

* Fix issue when barcode list argument contains non existing barcode(s) in input data or when all existing barcodes are used.

* The channel occupancy of the flowcell graph code has been rewritten to use Plotly. Add all/pass/fail/fail ratio views. The flowcell graph can now also handle Flongle and PromethION flowcels in addition of standard MinION flowcell
* Add "Sequencing kit" and "Barcode kits" entries in the run statistics table in html report
* In the distribution of read lengths graph, add buttons to show base count distribution in linear/log modes.
* Fix scatterplot graph where the default max x-axis value was always the max value for fail reads
* Fix help links in demo report
* Fix the name of the "Device and software" and "Run statistics" sections

* In setup.py, set the developement status for ToulligQC as Production/Stable instead of Beta
* Add MANIFEST.in file to add resources files in PyPi package
* Fix error with latest versions of NumPy by add a missing int casting

* Fix duration computation
* For PHRED score distribution boxplots, remove unnecessary interpolation before creating boxplot
* Remove duplicated code for 1D and 1D2 in PHRED score distribution graphs
* Remove duplicated code for 1D and 1D2 in Correlation between read length and PHRED score graphs
* In read length distribution graphs, add buttons to switch between linear and log scale for xaxis
* Add minimal qscore threshold in the "Device and software information" report table
* In 1D/1D2 sequencing summary extractors, now replace NA values for barcode assignment by "unclassified". Print a warning message on console
* Update unit tests
* For read count histogram tables, replace "frequency" by "percent"
* Rename y-axis for "Distribution of read lengths" graphs to "Read count"
* Add new command line options to finely define output file paths
* Update the sigma value for gaussian filters when smoothing plots
* In correlation scatterplots, now ponderate the number of pass/fail spots by the pass/fail ratio when using interpolation