Cache GEE catalog to avoid repeated network requests

[mihiarc]

Summary

• Add @lru_cache decorator to cache GEE catalog fetch, avoiding repeated network requests
• Consolidate duplicated fetch_geojson() implementations into shared functions in common.py
• Add helper functions for common catalog lookups

Problem

The GEE catalog was being fetched from GitHub multiple times during normal usage:

1. On class instantiation (to populate dropdown)
2. On dropdown value change
3. On button click
4. Inside load_gee_as_image() to determine dataset type

This resulted in 3-4 network requests per extraction operation, causing:

• Slow widget initialization
• Unnecessary network overhead
• Potential GitHub rate limiting on heavy use

Solution

Added cached functions to common.py:

@lru_cache(maxsize=1)
def fetch_gee_catalog():
    """Fetch and cache the GEE catalog for the session lifetime."""
    ...

def get_gee_catalog_as_dict(key_field="title", value_field="id"):
    """Get catalog as dict for dropdown menus."""
    ...

def get_dataset_type(dataset_id):
    """Get dataset type (image, image_collection, etc)."""
    ...

def get_dataset_url(dataset_id):
    """Get documentation URL for a dataset."""
    ...

Files Changed

File	Changes
skiba/common.py	+92 lines - New cached catalog functions
skiba/point_extraction.py	Refactored to use common functions
skiba/aggregated_point_extraction.py	Refactored to use common functions
skiba/buffer_coordinates.py	Refactored to use common functions
skiba/buffer_and_sample.py	Refactored to use common functions
skiba/buffer_method.py	Refactored to use common functions
skiba/geojson_buffering.py	Refactored to use common functions

Net change: -125 lines (removed duplication)

Additional Improvements

• Added 30-second timeout to network requests
• Better error handling (catches Timeout exception)
• Also fixes date filtering bug in additional locations (same as Bug: Date filtering logic is inverted in load_gee_as_image() #75)

Fixes #76