Brie

===========================
* Support saving detection table to tsv file
* Add the exon start and stop positions in brie-count

===========================
* Tune the learning rate with multiple values
* For test model, the fitted parameters will be used as initials
* Support base model with full features or null feature
* For gene feature only, switch sigma into per cell base
* Add noise term in simulator
* A few minor bug fix
* Implement a Inv-Gamma prior distribution for sigma (not in use)

===========================
* Support to use minimum minor isoform frequency to fileter genes (default=0.001)
* Add pseudo count (default=0.01) for none-empty element in both unique counts
for more robust results
* Reduce the sample size for Monte Carlo Expectation (10 to 3) for computational
efficiency
* Restructure the arguments in brie-quant
* Initialise the example notebook on multiple sclerosis data

===========================
* Fix minor bugs in brei-count and brie-quant cli for compatibility

===========================
* Change the whole BRIE model from MCMC sampler (v1) to variational inference (v2)
* Change the usage of each read to a summarised read counts for speedup
* Support splicing quantification without any feature or cell features or gene
features or both types of features.
* Support detection of variable splicing event associated with cell features
* Support acceleration with Graphic card (Nvidia GPU)
* Compatible with Scanpy analysis suite with a variety of plotting functions
* Restructure the whole package
* BRIE earlier version is still avaible with `brie1`

===========================
* fix a bug that fragment length longer than transcript length