Methylcheck

- Reading IDATs loading bar didn't work correctly, showed up after loading.
- Fixed error/logging messages:
- exclude_sex_control_probes() said 916 control probes were removed, then said “it appears your sample had no control probes”
- Erroneous message about missing values in poobah file: "color coding may be inaccurate."
- Filtering probes info message said there were N samples when it meant probes.
- methylprep.download.build_composite_dataset() Process time was negative.
- Target Removal and Staining graphs in plot_controls() had unreadable X-axis sample names. Labels are suppressed when
showing more than 30 samples.
- methylcheck.detect_array() sometimes returned array types in wrong case. All functions expect lowercase array types now.
- resolves exclude_sex_control_probes bugs.
- run_qc() and get_sex() did not recognize poobah_values.pkl on MacOS when using "~" in the filepath.
- methylcheck.problem_probe_reasons() lists probes matching any/all criteria when passing in no arguments, as documented
- get_sex() understands samplesheet ‘m’ and ‘f’ when not capitalized now.
- Load_both: always returns dataframe with probes in rows now, like .load() does.
- plot_M_vs_U now loads the noob_meth_values.pkl files if noob=True and files are found; otherwise it uses whatever meth/unmeth data is available.
- Methylcheck.qc_plot.qc_signal_intensity returns a dictionary of data about good/bad samples based on signal intensity.
Previously it was only returning this if 'plot' was False.
- controls_report() bug fixed: methylprep was producing samplesheet meta data pickles that contained Sample_ID twice,
because the GEO series_matrix files had this data appear twice. This broke the report, but this case is caught and avoided now.
controls_report() will recognize a wider array of samplesheet filenames now; anything with 'samplesheet' or 'meta_data' in the filename.

- added 'methylcheck report' CLI option to create a ReportPDF
- updated documentation
- minor bug fixes in read_geo()
- qc_plot() now handles mouse probe type differently
- handles importing from multiple pandas versions correctly
- read_geo can open series_matrix.txt files now

- fixed big where csv data_files were not included in pypi

- Improved ReportPDF custom tables option
- if fields are too long, it will truncate them or auto scale the font size smaller to fit on page.

- added GCT score to controls_report() used in the ReportPDF class.
- ReportPDF changes
- uses noob_meth/unmeth instead of raw, uncorrected meth/unmeth values for GCT and U vs M plot
- inverted poobah table to report percent passing (instead of failing) probes per sample
- this changed input from 'poobah_max_percent' (default 5%) to 'poobah_min_percent', (default 80%)
- M_vs_U not included by default, because redundant with qc_signal_intensity
- M_vs_U compare=True now labels each sample and has legend, so you can see effect of NOOB+dye correction on batch
- added poobah color-coding to MDS plot
- get_sex improved plotting
- will read poobah data and size sample points according to percent of failed probes
- save plots, or return fig, and more options now

- Added a controls_report() function that creates a spreadsheet summary of control probe performance.
- New unit test coverage. Note that because methylprep v1.4.0 changes processing, the results will change slightly
to match `sesame` instead of `minfi`, with nonlinear-dye-bias correction and infer-type-I-probe-switching.
- changed org name from FoxoBioScience to FoxoTech