Feature: Add merge function for combining rasters with a configurable aggregation

[b-j-karl]

Description

Add a high-level function to merge multiple rasters into a single output raster, using a user-specified aggregation function (e.g. minimum, maximum, mean) to resolve overlapping cells.

Motivation

Merging rasters is a common GIS workflow — e.g. combining tiled survey data or overlapping model outputs. Currently users must manually compute the union bounds, create a canvas, align each raster, and aggregate values cell-by-cell. A built-in function would make this straightforward.

Proposed Solution

A merge_rasters function in create.py with a signature along the lines of:

def merge_rasters(
    rasters: Collection[Raster],
    *,
    agg: Literal["min", "max", "mean", "sum"] = "min",
) -> Raster:

Behaviour:

• Compute the union of all input bounding boxes to define the output extent.
• Use the cell size and CRS from the first raster (all inputs must share the same CRS and cell size).
• Stack aligned arrays and apply the aggregation function, treating NaN as missing (i.e. np.nanmin, np.nanmax, etc.).
• Return a single Raster covering the full extent.

Alternatives Considered

• A Raster.merge() instance method — less natural since merging is N-to-1, not a transformation of a single raster.
• Accepting an arbitrary callable for agg — could be added later, but a Literal set covers the most common cases and keeps validation simple.

Additional Context

Fits naturally in create.py alongside other raster-construction functions like full_raster and rasterize_gdf.

[nathanjmcdougall]

We already have something like this in ttpy-geode I think.