Piecewiserefactor

[tsostarics]

This PR refactors the piecewise_interpolate_pulses function and its related functions (average_pitchtracks, interpolate_equal_pulses, etc.). See discussion in #10 for details, but the memory footprint and speed is drastically improved. In the process, a few major changes have been made:

• Parallelization capability is deprecated, if you want to parallelize the interpolation/averaging you should batch your dataset into some desired number of chunks. The parameter is included to avoid breaking previous analyses that used this.
• Index column now inherits the values from section_by if set to NULL
• Guessing behavior for indices has been removed, which eliminates some stubborn notes from Rcpp related to the methods for that function.
• Documentation has been updated for the index column
• .sort now defaults to FALSE, since F0 datasets are typically already sorted.

Under the hood behaviors that have changed:

• Copy isn't made with data.table anymore
• A larger focus on base R where possible
• Less flexibility with grouping structures; instead of trying to juggle user-set groupings and groupings from .grouping and aggregate_by, we just return an ungrouped dataframe. This avoids having to keep regrouping things, which adds up quickly.
• Instances where the dataframe was split into a list of dataframes has been removed, as this had a huge memory footprint and could result in the computer running out of memory if the dataset was too large and too many parallel calculations were being run
• There's a dependency on reframe now rather than working with summarize, may revisit this later if I find out it breaks something of mine

this closes #10