add emergency-stop.sh to halt a running migration

[DmitriiAn]

What

Adds pgcopydb-helpers/emergency-stop.sh — a one-shot emergency halt for a
running migration

Why

There was no safe, single command to stop a migration in an emergency (e.g. the
source DB is overloaded during the initial copy). Users had to find PIDs and
kill them by hand under pressure.

Behavior

• Confirms first, printing the consequences and live progress/phase read
  straight from the pgcopydb SQLite catalog.
• Stops the whole process tree: SIGTERM the process group (pgcopydb is its own
  group leader via setpgrp), then escalate automatically to repeated SIGKILL.
• Never leaves pgcopydb running silently: if anything survives SIGTERM + repeated
  SIGKILL, it prints a loud warning with the surviving PIDs and manual
  remediation, and exits 1.
• Stop-only: does not drop the replication slot or touch the target, so the
  migration stays resumable (resume-cdc.sh if the initial COPY finished, else
  resume-migration.sh). Prints next-steps for resume vs. abandon.

Safety

• Matches pgcopydb by exact process name (pgrep -x); never prints command lines
  (avoids leaking PGURI passwords); requires no ~/.env.
• Guard ensures it never signals its own process group.