Eflowcalc

* Algorithm Enhancements:
* checks for the format of the keyword argument 'hydro_year' in Calculator to make sure it is a valid 2-digit day/2-digit month format
* adds a conditional statement on the return of Calculator to send back the output array in the same orientation as the input streamflow array
* for dl19: deals with zero divide issues and returns zeros in place of NaNs when the denominator is null
* for fl3: deals with empty slice issues and replaces NaN values by zeros when the slice is empty and avoids raising unnecessary warnings
* for count_reversal: deals with special case where the null difference happens at the very beginning of the period, where NaN values were forward replaced (i.e. null difference) as long as the difference remained null. Now, after forward replacing, it backward replaces these NaN values for this special case at the beginning of the period if necessary

* Bug fixes:
* removes hard coded month and year of start of hydrological year that prevented Calculator from honouring the use of its keyword argument 'hydro_year' (if a value different from '01/10' was required)
* for count_reversal function: corrects error on the array index used to replace null difference by the most recent non-null value (low likelihood of having produced numerically erroneous outputs without running into an exception first with previous versions of EFlowCalc though)

* Documentation fixes:
* updates the tutorial notebook to correct typos

* Semantics Enhancements:
* adds 142 streamflow characteristics of the 171 inventoried by [Olden and Poff (2003)](https://doi.org/10.1002/rra.700)

* Functionality Enhancements:
* adds possibility to select specific years where to calculate the SFCs using the keyword argument `years`
* adds gathering functions for each flow category (`magnitude`, `frequency`, `timing`, `duration`, `rate_change`) and for all SFCs (`everything`)

* Algorithm Enhancements:
* adds single-sourcing package version
* changes default array orientation
* changes the degrees of freedom to n-1 in calculation of std (breaking backwards compatibility)