[HOLD] 0.10.0: runtime 0.98.0 pin + host prices on the map (host-pricing-v1)

[pietechsolution]

Ready for the batched SDK cut: runtime 0.99.0 is live on the CDN. The coordinator merges, tags v0.10.0 and publishes; the tag triggers publish.yml.

What changes

1. Runtime pin 0.84.1 → 0.98.0 (177b6e3). The phone map gets the current web picker's section overview, seat rendering and row labels. The bridge protocol range is unchanged (1–2); the only bridge changes between the two runtimes are additive (maintenance-pause message; seats added after checkout listed in the runtime snapshot).
2. Host prices reach the map (fae7d6c). SeatLayerPickerOptions.pricing.categories was declared but never sent, so a host selling from its own catalogue saw the chart's prices on the rail, the seat card and the cart. The picker now boots the runtime with them (init.config.pricing), and SeatLayerPickerController.setPricing changes them without a reload. Needs a runtime advertising host-pricing-v1; an older runtime keeps the chart's prices. Display only: the hold is still priced by SeatLayer. picker-spec.md §5 lists the capability for the iOS/Android/React Native ports (parity is a follow-up, not in this PR).
3. Version 0.10.0 (45bd9f2).

Before publishing

• Move seatLayerHostedWebVersion from 0.98.0 to 0.99.0 (the first runtime with host-pricing-v1) and update test/payloads_test.dart. Done; release.json for 0.99.0 is promotable: true.

Validation

• flutter analyze: no issues. bash scripts/check-public-repository.sh: passed.
• New test/picker_host_pricing_test.dart (wire shape, boot payload, stable profile): pass. file_size_test, package_version_test, payloads_test: pass.
• Real device run (iOS simulator, a DKK test event whose chart still carries EUR-era prices), with a pre-release runtime build that has host-pricing-v1:
  • host prices passed: rail DKK 935 / 673 / 560 / 486, seat card Gold · DKK 486, cart line and total DKK 486, host checkout DKK 486,00;
  • no prices passed (previous app build, same runtime): rail stays on the chart's DKK 125 / 90 / 75 / 65.

Pre-existing golden failures (not from this PR)

flutter test shows 66 failures, all golden pixel comparisons (e.g. goldens/dock_bar_light.png, 0.37% diff) on this machine. Untouched main (f7b25e9) fails the identical set of 66 tests on the same machine. I diffed the two failure lists and they match line for line, so this is a renderer/toolchain difference, not this change. The one extra failure this branch introduced (file_size_test, controller at 1502 lines) is fixed (1499).