feat(cli): add tcping subcommand

[Kelvinoppong]

Summary

Adds a new kubectl retina tcping HOST PORT subcommand that probes a TCP port by performing a full connect/close handshake, reports per-attempt RTT, and prints min/avg/max/loss statistics at the end (or on Ctrl+C).

This resolves #2022, which asked for a simple tool to generate continuous TCP traffic for debugging without needing curl in a while loop or heavier tools like nmap.

Behavior

• Runs continuously until interrupted (SIGINT/SIGTERM) or until --count probes are sent.
• Supports IPv4, IPv6, and hostnames (uses net.JoinHostPort + net.DialTimeout).
• Prints a summary on exit including partial stats when interrupted mid-run.

Flags

Flag	Default	Description
-c, --count	0 (unlimited)	Number of probes to send
-i, --interval	1s	Interval between probes
-t, --timeout	2s	TCP connect timeout per probe

Example

$ kubectl retina tcping 1.1.1.1 443 --count 3
TCPing 1.1.1.1:443: tcp connect ...
seq=1 1.1.1.1:443 rtt=9.893ms
seq=2 1.1.1.1:443 rtt=8.576ms
seq=3 1.1.1.1:443 rtt=11.077ms

--- 1.1.1.1:443 tcping statistics ---
3 probes sent, 3 successful, 0.0% loss
rtt min/avg/max = 8.576ms/9.849ms/11.077ms

Test plan

• Unit tests covering success, connection failure, stats aggregation (min/max/avg/loss), help output, and arg validation (cli/cmd/tcping_test.go).
• All existing CLI tests still pass (go test ./cli/...).
• go vet ./cli/... clean.
• Manual smoke tests:
  • IPv4 happy path (1.1.1.1:443)
  • Hostname/DNS resolution (google.com:443)
  • IPv6 ([::1]:22)
  • DNS failure (nonexistent.invalid.domain:80)
  • Connection refused (127.0.0.1:1)
  • SIGI summary with exit 0

Fixes #2022

[Kelvinoppong]

@nddq I have tackled another issue and this PR is ready for review.