Support tables outside the public schema

[lukashes]

Closes #50.

Streams could only target the public schema: startup validation hardcoded it and matchStreams dropped any change from another schema, even when the table name collided.

• resource is a fully-qualified name (schema.table) end to end. The Postgres converter emits it and matchStreams compares it as an opaque string, so routing stays source-agnostic (no schema concept in the domain).
• The config accepts a bare name and normalizes it to public.<name>; write schema.table to target another schema.
• Breaking (message format): meta.schema is removed and meta.resource now carries the qualified name (public.users, was users). v0, noted in the changelog.

[github-actions]

📊 Benchmark Results

Current run is the minimum over 3 passes, compared against the base branch (main @ eea6c0c), built on the same runner.

Benchmark	Baseline	Current	Δ Time	Allocs	Status
Converter DELETE	134.56μs	152.31μs	+13.2%	14 → 13	🔴 slower
Converter INSERT	135.21μs	152.60μs	+12.9%	14 → 13	🔴 slower
Converter UPDATE	183.82μs	184.47μs	+0.3%	23 → 22	➡️
JsonSerializer	53.44μs	53.72μs	+0.5%	3 → 3	➡️
KafkaProducer produce	211.82μs	202.49μs	-4.4%	1 → 1	➡️
KafkaProducer send	257.42μs	257.44μs	+0.0%	0 → 0	➡️
PgOutputDecoder	94.37μs	102.89μs	+9.0%	6 → 6	🔴 slower
getPartitionKeyValue boolean	17.54μs	16.80μs	-4.2%	1 → 1	➡️
getPartitionKeyValue integer	0.04μs	0.04μs	+0.0%	0 → 0	⚪ noise
getPartitionKeyValue not found	0.04μs	0.04μs	+0.0%	0 → 0	⚪ noise
getPartitionKeyValue string	19.00μs	18.84μs	-0.8%	1 → 1	➡️
matchStreams found	19.14μs	18.96μs	-0.9%	1 → 1	➡️
matchStreams not found	0.06μs	0.06μs	-3.4%	0 → 0	⚪ noise

Summary: ➡️ 7 neutral · 🔴 3 slower · ⚪ 3 ignored (sub-μs)

_{Thresholds: <1μs ignore · 1–20μs 15% · 20–50μs 10% · ≥50μs 5%. Measured on a shared CI runner — treat small deltas as noise. Informational only; this check never fails the build.}

[lukashes]

/bench

[lukashes]

/bench

[lukashes]

/bench