Parallelize support bundle collection and add Neo4j diagnostics

[josho-sysdig]

Refactor get_support_bundle.sh to dispatch independent collection tasks as concurrent background jobs, reducing wall time on representative clusters (~40 pods, ~120 containers) from ~10m 36s to ~3m 32s with the default MAX_JOBS=6 setting.

Parallelization changes:

• Introduce three job-control helpers: run_bg (fork+register), throttle (cap concurrency via wait -n / sleep 0.1 fallback for Bash <4.3), and wait_all (harvest PIDs, report failures, clean up temp output files)
• Add MAX_JOBS variable (default 6, overridable via --max-jobs flag or environment variable) to prevent API server overload
• Refactor container log + support-file collection into collect_container_logs(), launched one background job per container
• Refactor per-node JSON manifest collection into collect_node_info(), launched one background job per node
• Refactor per-resource-type manifest collection into collect_resource_manifests(), launched one background job per type
• Refactor Cassandra stats/storage, Elasticsearch stats, PostgreSQL, MySQL, Kafka, and Zookeeper storage into dedicated functions, each dispatched as a background job
• Move kubectl cluster-info dump to a background job so it runs concurrently with log collection
• Directory creation and pod/node/container discovery remain serial to avoid race conditions

New collection: Neo4j diagnostics

• Add collect_neo4j_stats() to retrieve cluster server and database status via cypher-shell, reading the ingestion_admin password from the neo4jdb-user-secrets or neo4j-user-secrets Kubernetes secret
• Outputs neo4j//cypher_show_servers.txt and neo4j//cypher_show_databases.txt; writes a skip log if the secret is unavailable

README.md rewrite:

• Replace minimal usage stub with full man-page-style documentation covering NAME, SYNOPSIS, DESCRIPTION, OPTIONS, ENVIRONMENT, OUTPUT (annotated directory tree), EXAMPLES, and PARALLEL PROCESSING sections
• Document all flags including the new --max-jobs option
• Add neo4j/ output directory to the archive structure reference
• Include concurrency control design notes and performance benchmark table