fix: follow up to #45

[simonmysun]

No description provided.

[Copilot]

Pull request overview

This PR is a follow-up to #45 that tightens up the shell implementation and accompanying docs: it clarifies installation via a stable ell launcher, makes plugin hook handling more robust, improves error reporting for backend HTTP failures, and hardens JSON parsing.

Changes:

• Document an alternative install method by symlinking the ell launcher into ~/.local/bin (and explain why to link ell, not ell.sh).
• Improve shell robustness by scoping variables with local, reading plugin hook lists into arrays via mapfile, and enhancing curl failure diagnostics.
• Harden JSON number parsing by validating scanned number tokens against RFC 8259 grammar.

Reviewed changes

Copilot reviewed 10 out of 10 changed files in this pull request and generated no comments.

Show a summary per file

File	Description
README.md	Adds symlink-based install guidance referencing the ell launcher.
plugins/paginator/90_pre_output.sh	Adds local scoping for cursor-position helper variables.
llm_backends/openai/generate_completion.sh	Captures curl exit status for clearer fatal errors; returns non-zero on JSON parse failure.
llm_backends/gemini/generate_completion.sh	Same curl-status handling and JSON-parse failure behavior as OpenAI backend.
helpers/piping.sh	Scopes the internal pipeline command string with local.
helpers/parse_arguments.sh	Scopes parsing temporaries with local.
helpers/load_config.sh	Scopes config loader temporaries with local.
helpers/json.sh	Validates JSON number tokens against RFC 8259 number grammar.
ell.sh	Uses mapfile to treat plugin hook lists as arrays and improves related debug output.
docs/Configuration.md	Updates documented default ELL_TEMPLATE to default-openai to match the script.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.