[Bug]: jira workitem get --full hangs indefinitely; limited --fields get returns in seconds

## What happened

`twg jira workitem get --full -o json` did not return within 120 seconds and had to be backgrounded. Its stdout file stayed empty, and it was still unfinished when the session ended some time later. The same issue key with an explicit field list returned in a few seconds:

```bash
# hangs (>120s, empty output)
twg jira workitem get SLP-178 --full -o json

# returns in seconds
twg jira workitem get SLP-178 --fields summary,description,status,issuetype,priority,labels,parent -o json
```

## What I expected

`--full` returns, or fails with an error, well inside a normal timeout — or at minimum emits partial output or progress so a caller can tell it is still working rather than wedged.

## Why it matters

`--full` is the documented one-shot read (`Fetch full issue fields, complete comments, and Jira remote links`). A script or agent that reaches for it first gets no output and no error, and cannot distinguish a hang from a slow network. The workaround — enumerating fields by hand — requires already knowing which fields you want, which is the thing `--full` exists to avoid.

## Steps to reproduce

1. Authenticate against a Jira Cloud site.
2. Run `twg jira workitem get --full -o json` on an issue whose description contains an embedded media node and which has at least one label.
3. Observe no output for 120s+.
4. Run the same key with `--fields summary,description,status,issuetype,priority,labels,parent -o json` and observe a normal fast response.

## Environment

- `twg` version: **1.2.7**
- OS: Windows 11 Enterprise 10.0.26200
- Shell: Git Bash (also reproduced from a PowerShell-primary environment)
- Site: Jira Cloud (`*.atlassian.net`)
- Issue used: standard Task, 2 labels, ADF description with one `mediaSingle` node, no parent

## Additional notes

Unverified, listed only as a lead: `--full` fans out to supplemental requests for comments and remote links, so one of those calls may be the one stalling rather than the base field read. Either way, a per-request timeout with a clear error would be preferable to an indefinite wait.

## Possibly separate, smaller issue

`twg jira workitem get --id SLP-178 ...` fails with `error: unknown option '--id'`, while `twg help describe "jira workitem get"` documents `--issue-id` and `--key` as aliases for the positional key. Some agent-facing skill docs use `--id`. Either the alias is missing or the docs overstate it — happy to split this into its own issue if you prefer.

Start by running `twg jira workitem get --full -o json` and compare it with the explicit `--fields` invocation. Trace the `--full` path, including its comments and remote-links requests; done means the command returns or fails with a clear error within a normal timeout instead of hanging indefinitely.

Written by the indexing model from the issue text.