Resolves #602 GlacioHack/xdem#660

Description:

This pull request introduces enhancements to the statistics calculation methods in the project. The goal is to provide users with more flexible and comprehensive statistics functionality for raster data. The get_stats method has been created to:

• Return a list of statistics in a dict containing mean, median, max, mean, sum, sum of squares, 90th percentile, nmad, rmse, std.
• Support single statistic: User can request one statistic.
• Support multiple statistics: Users can request multiple statistics in one call.
• Support custom callables: Users can pass any callable function to calculate custom statistics.

A list of aliases has been drawn up in order to be as flexible as possible when users request statistics.

Changes:

• NMAD in geoutils.stats
  note : nmad can't be called directly from a raster because of circular imports, you have to go through an array like.
• _statistics method: Calculates common statistics (mean, median, max, min, sum, RMSE, etc.) for a specified band of raster data.
• get_stats method: Retrieves requested statistics by their name (either from a predefined list or user-defined), with support for passing a callable function.
• _get_single_stat method: Handles fetching a single statistic by matching it with its alias or function name. Prints a warning and return NaN if the statistic required is not in the list
• Support for callable statistics: Users can now provide any custom function that reduces the raster data (such as custom percentiles).

Usage Examples:

1. Get all statistics:

   stats = raster.get_stats()

2. Get a single statistic (e.g., 'mean'):

   mean_value = raster.get_stats("mean")

3. Get multiple statistics:

   selected_stats = raster.get_stats(["mean", "max", "rmse"])

4. Using a custom callable statistic:

   def custom_stat(data):
       return np.nansum(data > 100)  # Count the number of pixels above 100
   custom_value = raster.get_stats(custom_stat)

Tests

The following tests ensure that the new functionality in the get_stats method works as expected:

1. Full Statistics:

   • The test retrieves all available statistics (mean, median, max, min, sum, etc.) for a raster and ensures each statistic is present and not None.

2. Single Statistic:

   • A test is included to fetch a single statistic, "Average" (which is an alias for "Mean"), and verifies that the returned value is a float.

3. Selected Statistics:

   • The test checks the retrieval of a selected set of statistics: mean, maximum, and std. Additionally, it includes the use of a custom statistic function (percentile_95), which computes the 95th percentile of the data.
   • This validates that both built-in and user-defined statistics can be fetched successfully.

4. Unrecognized Statistics:

   • Ensure that get_stats return NaN and a warning is raised if the statistic requested by the users is not in the list of aliases

5. NMAD tests in test_stats:

   • Ensure the good behavior of the scale factor
   • Check if the returned nmad is the same than scipy.stats.median_abs_deviation

Documentation

A section about how to use get_stats has been added to the raster class documentation.