update dependencies and enhance connection handling in metrics

[OutOfBedlam]

This pull request introduces several improvements and fixes across the connection pooling, metrics collection, and SQL driver testing subsystems. The most significant changes include making the metrics collector non-blocking under load, improving connection pool error handling and configuration, and expanding test coverage for SQL wrappers and connection pooling.

Metrics Collection Improvements:

• The Collector in metric.go now uses a much larger buffer (recvChSize increased from 100 to 10,000) and is non-blocking: if the buffer is full, new measurements are dropped and a counter is incremented to track how many were dropped. This prevents request goroutines from blocking due to backpressure. A new method, DroppedCount(), exposes the number of dropped measurements. Unit tests verify this behavior. [1] [2] [3] [4] [5]

Connection Pooling and Error Handling:

• The default database pool is now configured with sensible limits for open/idle connections and connection lifetimes, and is proactively pinged to catch errors early.
• The HTTP query handler now strictly requires a working connection pool; if unavailable, it returns HTTP 503 (Service Unavailable) with a Retry-After header, rather than falling back to a direct connection.
• System metrics no longer report pool availability as a binary metric, but continue to collect pool stats if available.
• The dashboard exposes new charts for connection pool stats, including open/in-use/idle connections, wait durations, and closed connections.

Testing and SQL Wrapper Enhancements:

• Test stubs for pooled connections are improved for more realistic behavior, and tests are updated to check for proper error handling and connection counting. [1] [2] [3] [4]
• Additional tests cover edge cases and error branches in SQL row and connection wrappers, including type mapping and error propagation. [1] [2] [3] [4] [5]

Dependency Update:

• The github.com/machbase/neo-client dependency is updated to a newer version.

These changes collectively improve system robustness under load, provide better observability for connection pooling, and increase test reliability and coverage.

[codecov]

Codecov Report

❌ Patch coverage is 46.51163% with 23 lines in your changes missing coverage. Please review.
✅ Project coverage is 65.46%. Comparing base (2a57ab6) to head (350e6e5).

Files with missing lines	Patch %	Lines
mods/server/http_query.go	17.64%	12 Missing and 2 partials ⚠️
mods/util/metric/metric.go	66.66%	3 Missing and 3 partials ⚠️
spi/metrics.go	0.00%	3 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #433      +/-   ##
==========================================
+ Coverage   65.40%   65.46%   +0.05%     
==========================================
  Files         351      351              
  Lines       69696    69720      +24     
==========================================
+ Hits        45585    45640      +55     
+ Misses      19264    19229      -35     
- Partials     4847     4851       +4     

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.