Miscellaneous bug fixes

[warrenmcg]

Hi @pimentel and @lynnyi,

I have completed the to-do list I created in #213, but figured I should open a separate pull request to incorporate them all. There might be an issue with my shiny changes (6ff28d2 and
193745f), but otherwise, everything else should be solid.

Here is a list of issues that have been addressed:

• likelihood test table in shiny has no down button #62 added download buttons for the LRT test table and for the fragment length distribution plots (everything else appeared to have download buttons)
• Unstated namespace dependency on data.table::head #135 fixed dependency issue with utils::head
• Issue reading h5 files #120 and HDF5 files not being read properly (hdf5 lib incompatibility, or base R and rhdf5 clash?) #175 provide a more informative error message if a valid HDF5 file cannot be read, as this appears to be the most common reason why
• add support of strings as factors #197 (coercing kallisto paths to character if they have been provided as a factor)
• subsetting kallisto object using subset_kallisto()  #204 (subset_kallisto was not subsetting the abundance matrix correctly)
• fixed plot_cluster_heatmap bugs found in Order of labels_row in plot_transcript_heatmap #207 and cluster_transcripts = FALSE still cluster rows in plot_transcript_heatmap #208.
• typo in ?sleuth_prep #212 fixed the typo found in the documentation (% needs to be escaped)

[johanneskoester]

Any chance to get this merged and a new release? The heatmap bug can lead to severe misinterpretations if people faithfully try to replace the row labels.