DotNetWorkQueue v0.11.0

What's Changed

• perf(core): stop allocating in argument validation and the body-type header by @blehnen in #230
• feat(#236): opt-in System.Text.Json serializer, selected per message by @blehnen in #237
• perf(litedb): scope the send and receive locks to a database, not the process by @blehnen in #238
• perf(litedb): send a batch in one transaction instead of a parallel loop by @blehnen in #239
• docs(litedb): correct the shared-connection finding, and measure the batch there by @blehnen in #240
• perf(litedb): stop scanning the queue to find the next message by @blehnen in #241
• perf(core): cut what every send allocates, and fix the Memory batch ordering (#235) by @blehnen in #242
• perf(core): take the consume path out of the DI container (#235) by @blehnen in #244
• test(benchmarks): close out the LiteDb pass — concurrency and allocation, both falsified (#234) by @blehnen in #245
• perf(sqlserver): cache the de-queue statement for routed consumers, and the meta insert SQL (#231) by @blehnen in #246
• perf(postgresql): cache the de-queue statement for routed consumers (#232) by @blehnen in #247
• docs(postgresql): auto-prepare measured and documented, deliberately not enabled (#232) by @blehnen in #248
• test(benchmarks): the relational decorator stack is not a lever (#231, #232) by @blehnen in #249
• SQL Server: send an ordinary message in one round trip (#231) by @blehnen in #250
• PostgreSQL: send an ordinary message in one round trip (#232) by @blehnen in #251
• Give the changelog rule a shape it can be held to by @blehnen in #254
• Parameterise the delay and expiration values in the meta insert by @blehnen in #257
• Redis performance pass: measure both paths, no change warranted by @blehnen in #259
• Sonar cleanup after the performance sweep by @blehnen in #261
• Run the Dashboard E2E tests as a non-root agent by @blehnen in #263
• Prepare 0.11.0: rewrite the changelog for its readers by @blehnen in #262

Full Changelog: v0.10.0...v0.11.0