Give each Draw export button a unique element id

[Htet-Kaung-San]

The Draw export button is hard-coded as id='export', and its click handler looks it up with document.getElementById('export'). Adding two Draw controls with export=True emits two elements sharing one id — invalid HTML, and the feature breaks:

import folium
from folium.plugins import Draw

m = folium.Map()
Draw(export=True, filename="a.geojson").add_to(m)
Draw(export=True, filename="b.geojson").add_to(m)
m.save("out.html")

On main that renders two <a id='export'> elements. Every getElementById('export') call resolves to the first one, so the second onclick assignment overwrites the first. The result is that the first button exports the second control's FeatureGroup under the second control's filename, and the second button does nothing at all.

This derives the id from get_name() instead, which is what the other plugins injecting their own elements already do — ScrollZoomToggler and FloatImage both use id="{{this.get_name()}}".

I also dropped the duplicated top: 5px from the button's CSS. top: 90px is declared later in the same block and already won, so it was dead code — the rendered position is unchanged.

On the id change

The id is not documented as public API, but anyone currently styling the button with a #export { ... } rule would need to switch to targeting the element another way. Happy to keep a stable class alongside the unique id if you'd rather not break that.

I found this while reading the plugin for #1806; this PR doesn't address that request.

Testing

tests/plugins/test_draw.py is new — the Draw plugin had no test coverage. It covers the unique id, the export=False case, and two controls each staying wired to their own FeatureGroup and filename. The two id-related tests fail on main and pass here.

Full suite, excluding tests/selenium and tests/snapshots (missing pixelmatch locally): 258 passed, 6 failed. The same 6 fail identically on an unmodified checkout — test_timedynamic_geo_json, test_dead_map_id_rule_absent, test_icon_invalid_marker_colors, and three test_repr.py png tests — so none are from this change.

ruff and codespell are clean. black wants to rewrap the Template(...) call in draw.py, but it asks for exactly that same rewrap on an unmodified checkout; I get black 25.11.0 locally since 26.5.1 doesn't install on Python 3.9, so I left the formatting alone rather than add unrelated churn.

[adarshsm]

Nice catch and a clean fix. I went through it against main and it holds up:

• Root cause confirmed: two Draw(export=True) controls emit two <a id='export'> elements, and getElementById('export') only ever resolves the first — so the second button is dead and the first exports the wrong FeatureGroup. Deriving the id from get_name() matches how ScrollZoomToggler/FloatImage already namespace their elements.
• The top: 5px removal is safe: it and top: 90px sit in the same #export {} rule, so top: 90px already won — rendered position is unchanged.
• Tests: ran tests/plugins/test_draw.py locally, all 3 pass including the two-controls regression, and confirmed the emitted ids are unique.

One small thing worth a mention (not a blocker): this renames the element id from export to export_<name>, so anyone with custom CSS targeting #export would need to update it — fine given it was duplicate/invalid HTML anyway, just flagging in case it's worth a release note.

LGTM.