Bash comm/join/sort Set Reconciliation Prompt

You are a senior SRE. Write a portable bash script that reconciles two flat lists of identifiers (hostnames, IDs, or keys) using only coreutils `sort`, `comm`, and `join` — no Python, no awk-heavy logic. Follow these steps:

1. Set strict mode (`set -euo pipefail`) and accept two input files as arguments: [LIST_A] (e.g. the cloud inventory export) and [LIST_B] (e.g. the CMDB or config-management host list).
2. Normalize each input into a temp file: strip blank lines and comments, trim whitespace, optionally lowercase, then `sort -u`. Both `comm` and `join` REQUIRE sorted input, so this step is mandatory and must use the same collation (`LC_ALL=C sort`) for both files.
3. Use `comm -23` to report identifiers present in [LIST_A] but not in [LIST_B] (orphans / drift), and `comm -13` for those present in [LIST_B] but not in [LIST_A] (missing).
4. Use `comm -12` (or `join`) to report the intersection — identifiers present in both.
5. If a second column carries metadata (e.g. `host status`), use `join -j 1` on the sorted files to correlate that metadata across both sources for the intersection set.
6. Emit a summary line with the counts for each of the three sets, and write the three detail lists to clearly labeled sections or to [OUTPUT_DIR] as `only-a.txt`, `only-b.txt`, and `both.txt`.

Output format: a human-readable report to stdout with three labeled sections (Only in A / Only in B / In both) and a trailing counts summary; optionally machine-readable files when [OUTPUT_DIR] is set.

Guardrail: the script is strictly read-only — it must never modify, delete, or reconcile the actual inventory; it only reports the diff. Use `mktemp` for all intermediate files and clean them up with a trap so repeated runs leave no residue and are fully idempotent.