Description

This PR closes #742

742 mentions that `pit_sample()` currently fails when there is only a single observation. This shouldn't be the case.

This PR

sparks joy
simplifies input checks for pit_sample()
makes it possible to compute a single PIT value (by converting the input from a vector to a matrix with one row)
adds a new unit test and updates existing tests to reflect the changes in the input checks
Adds a seed to the plot_pit() snapshots. The discrete pit is randomised and I think making calls to rnorm changes the state of the random number generator. That means that if you add an rnorm() call to previous tests it can change the snapshot. That happened to me once before and I hope this fixes it.
drops the function assert_not_null() and its tests, as the function is no longer used
simplifies the process of checking whether forecasts are discrete or continuous by using the internal get_type() function
has no unrelated changes. Am I still me? It seems like scoringutils has changed me in unexpected ways....

[x] My PR is based on a package issue and I have explicitly linked it.
[x] I have included the target issue or issues in the PR title as follows: issue-number: PR title
[x] I have tested my changes locally.
[x] I have added or updated unit tests where necessary.
[ ] I have updated the documentation if required.
[x] I have built the package locally and run rebuilt docs using roxygen2.
[x] My code follows the established coding standards and I have run lintr::lint_package() to check for style issues introduced by my changes.
[ ] I have added a news item linked to this PR.
[x] I have reviewed CI checks for this PR and addressed them as far as I am able.