Update Benchmark topic with latest valkey-benchmark.c changes

[Andriciuc]

This PR focuses on updating the benchmark.md chapter from #414 and using the latest updated valkey-benchmark.c file from valkey-io/valkey#2823.

First commit changes:

add --duration and --warmup options
move --seed in correct option place
add two examples based on --duration, --warmup and --rps

Replaces #435

[Andriciuc]

The following changes have been done to the benchmarking chapter:

For Placeholders (Description section):

• rand_int: Added note that multiple occurrences within a single command each get different values
• Added rand_1st through rand_9th — same-value placeholders with independent counters
• Added field:name — dataset field substitution, requires --dataset

New options added:

• --duration — time-based benchmark run, mutually exclusive with -n
• --warmup — pre-benchmark warmup period, stats discarded
• --sequential — sequential key replacement modifier for -r
• --rps — request rate limiter using token bucket algorithm
• --rfr — read-from-replica mode for cluster benchmarks
• --num-functions — controls function_load test (default 10)
• --num-keys-in-fcall — controls fcall test (default 1)
• --dataset — CSV/TSV file for field placeholder substitution
• --maxdocs — limits records loaded from dataset
• --fuzz — random command generation (testing only)
• --fuzz-mode — fuzzing modes: malformed-commands, config-commands
• --fuzz-loglevel — fuzzer log verbosity
• --rdma — RDMA connection support
• --mptcp — Multipath TCP support

Existing option corrections:

• -r: "expand" → "replace"; added per-occurrence behavior note
• --precision: default corrected from 0 to 3

Examples:

• Added time-based benchmark example (--warmup, --duration)
• Added rate-limited benchmark example (--rps)
• Added dataset benchmark example (--dataset, field:name)
• Removed \n\n"); C source artifact from transaction example

Keyspace output block:

• Removed backtick artifacts around <= in the sample output