Make the download cache insensitive to region-name spelling - #7

Small, but helpfull change, making the caching case insensitive

The bug

The --country value is embedded verbatim in every cache key, so the spelling of the region name decides whether a cached download is found:

$ python create_grid_poster.py --country Germany # downloads 12 tiles, ~40 min
$ python create_grid_poster.py --country germany # downloads all 12 again
Using cached boundary for germany
Downloading OSM power features: power=['line'] across 12 tiles

Confirmed against a populated cache:

The boundary line is misleading — that geocode happened to be cached under the lower-case spelling from an earlier attempt — but the tiles are the expensive part, so the run re-queries Overpass in full. On a rate-limited public endpoint that is the difference between an instant render and a long throttled download. Stray surrounding whitespace splits the cache the same way.

The fix

Fold case and surrounding whitespace when building cache keys, via a single _cache_name() helper applied at the six sites where the region name feeds cache_key(). Nothing else changes — the name is still displayed and geocoded exactly as typed.

Verification

• "Germany", "germany", "GERMANY", " Germany " → 1 distinct cache key (was 4).
• Names differing by more than case stay distinct: Germany/France/North America/United States of America → 4 distinct keys.

Compatibility note

This changes existing key values. A cache written with a capitalized name is re-downloaded once; keys for names already typed in lower case are unchanged, since the normalized form equals the old raw form. Worth a mention in release notes if that matters for the gallery build.