impactr 0.4.1

Fixed an issue with saving the non-wear plot with remove_nonwear() (#2).
Change the name of the "valid_observation" to "valid_file" in the non-wear summary to better express its meaning.
Limited the number of characters of the non-wear plot title to 50 characters, preventing the plot title to exceed the plot window limits. In case of large (n. char. > 50) titles, remove_nonwear() automatically crops it.
Return NA in the summary variables from summarise_loading() whenever the number of detected peaks is 0.
Change the coefficients of the prediction models for walking/running to match the final version of the published paper.

impactr 0.4.0

Added the function remove_nonwear() to detect and remove accelerometer non-wear time.
Added the function summarise_loading().
Include an interface to access example datasets from the {accdata} package. Run ?import_dataset for help.
Changed how resultant vector is computed to improve speed.
read_acc() no longer displays a progress bar.

pracma::findpeaks() is now used to get the index of the curve start.
Fixed a bug in which predict_loading() did not return the expected columns if outcome is set to "all".
Added a new supported model: "jumping".

define_region() now works with multi-day data. See the updated documentation.
specify_parameters() and filter_acc() throw errors when called more than once on the same data. This prevents attributes being accidentally added on top of existing ones.
predict() throws an error when required attributes are missing.
Fixed a test failure with {tibble} release 3.1.4 (#1).