Move back to default price

[epompeii]

The stopgap implementation of handling multiple Stripe subscription prices (#690) is no longer needed now that we've moved over to the usage-based meters (#689).
This changeset removes the metrics prices and the preferred price selection.

This PR is in draft until the migration is complete at the end of all billing cycles.

[github-actions]

🤖 Claude Code Review

PR: #694
Base: devel
Head: u/ep/default-price
Commit: 051ef9af16f6aff8e7921cc217700fe9a23ff5ef

---

Here's my review:

---

PR Review: default_price (051ef9a)

Summary

This PR removes the metered billing migration scaffolding that was introduced in #689 (Stripe usage-based meters). It consolidates the "default" and "metrics" price entries into a single "default" entry per product tier and removes the filter_subscription_items / preferred_price_ids methods that were only needed during the transition period.

Assessment: Looks Good

This is a clean removal of temporary migration code. The changes are well-scoped and consistent.

Potential Concern

Risk of breakage if migration isn't complete: The filter_subscription_items method existed to handle subscriptions that still have both old and new price items. By removing it, get_subscription_item (biller.rs:568) now receives the raw subscription.items.data directly and will return BillingError::MultipleSubscriptionItems if a subscription still has more than one item. If any live Stripe subscriptions haven't been migrated yet (i.e., still have both the old "default" and new "metrics" subscription items), this will start erroring at runtime. Confirm that all existing subscriptions have been cleaned up in Stripe before deploying.

Code Quality

• Unused imports (HashSet, PriceId, DEFAULT_PRICE_NAME, METRICS_METER_EVENT_NAME) are properly removed
• The BillingError::NoMatchingSubscriptionItem variant is correctly removed from error.rs
• Tests are updated consistently: migration-specific filter tests deleted, remaining tests now use DEFAULT_PRICE_NAME instead of METRICS_METER_EVENT_NAME
• Test data simplified from two metered entries per product to one, matching the new config shape
• No dead code left behind

No Issues Found With

• Security: No concerns; this is internal billing plumbing with no new external inputs
• Performance: Slight improvement — removes an unnecessary HashSet allocation and filtering pass
• CLAUDE.md compliance: No lint suppressions added, thiserror used for errors, strong types throughout

---

_{Model: claude-opus-4-6}

[github-actions]

Bencher Report

Branch	u/ep/default-price
Testbed	ubuntu-22.04

Click to view all benchmark results

Benchmark	Latency	Benchmark Result microseconds (µs) (Result Δ%)	Upper Boundary microseconds (µs) (Limit %)
Adapter::Json	📈 view plot 🚷 view threshold	3.67 µs (+8.26%) Baseline: 3.39 µs	4.59 µs (79.91%)
Adapter::Magic (JSON)	📈 view plot 🚷 view threshold	3.66 µs (+8.13%) Baseline: 3.38 µs	4.51 µs (81.05%)
Adapter::Magic (Rust)	📈 view plot 🚷 view threshold	25.83 µs (+0.80%) Baseline: 25.62 µs	30.81 µs (83.83%)
Adapter::Rust	📈 view plot 🚷 view threshold	2.83 µs (-0.70%) Baseline: 2.85 µs	3.34 µs (84.52%)
Adapter::RustBench	📈 view plot 🚷 view threshold	2.86 µs (+0.50%) Baseline: 2.84 µs	3.32 µs (86.03%)
head_version_insert/batch/10	📈 view plot 🚷 view threshold	106.03 µs (+6.59%) Baseline: 99.48 µs	123.53 µs (85.83%)
head_version_insert/batch/100	📈 view plot 🚷 view threshold	242.58 µs (+2.65%) Baseline: 236.31 µs	269.29 µs (90.08%)
head_version_insert/batch/255	📈 view plot 🚷 view threshold	462.04 µs (+0.55%) Baseline: 459.51 µs	489.08 µs (94.47%)
head_version_insert/batch/50	📈 view plot 🚷 view threshold	163.81 µs (+2.59%) Baseline: 159.67 µs	185.38 µs (88.36%)
threshold_query/join/10	📈 view plot 🚷 view threshold	149.87 µs (+4.83%) Baseline: 142.97 µs	170.46 µs (87.92%)
threshold_query/join/20	📈 view plot 🚷 view threshold	165.51 µs (+5.36%) Baseline: 157.08 µs	185.47 µs (89.24%)
threshold_query/join/5	📈 view plot 🚷 view threshold	139.63 µs (+3.30%) Baseline: 135.16 µs	160.49 µs (87.00%)
threshold_query/join/50	📈 view plot 🚷 view threshold	201.61 µs (+1.84%) Baseline: 197.97 µs	228.38 µs (88.28%)

🐰 View full continuous benchmarking report in Bencher