[improve][broker] Improve dispatch performance by summing entry bytes with a loop

[void-ptr974]

Motivation

The shared-subscription dispatch path sums entry bytes for every read batch. The existing implementation uses a stream pipeline, which adds avoidable allocation and dispatch overhead on this hot path.

Modifications

Replace the stream-based byte summation in both multiple-consumer dispatchers with a simple indexed loop.

Performance

This improves the dispatch hot path by reducing per-read-batch CPU work and eliminating stream pipeline allocation.

JMH data gathered locally using DispatcherDispatchPathBenchmark.(streamTotalBytes|loopTotalBytes), with -p entriesCount=32,256,1000 -wi 2 -i 5 -w 1s -r 1s -f 1 -bm avgt -tu ns -prof gc:

entriesCount	streamTotalBytes	loopTotalBytes	improvement	allocation
32	112.4 ns/op	40.3 ns/op	64.1%	256 B/op -> ~0
256	407.3 ns/op	253.7 ns/op	37.7%	256 B/op -> ~0
1000	1886.0 ns/op	1565.8 ns/op	17.0%	256 B/op -> ~0

Verification

• Added PersistentDispatcherTotalBytesTest coverage for empty, single-entry, and large 1024-entry batches with varying real EntryImpl payload sizes.
• ./gradlew :pulsar-broker:compileJava :pulsar-broker:checkstyleMain --max-workers=1
• ./gradlew :pulsar-broker:test --tests org.apache.pulsar.broker.service.persistent.PersistentDispatcherTotalBytesTest --max-workers=1 -PtestRetryCount=0
• ./gradlew :pulsar-broker:checkstyleTest --max-workers=1