Refactoring of introspection

Nobody has claimed this yet.

## Brief feature description

The introspection needs a refactoring regarding:
- how is data internally stored and accessible for introspection
- how is this data sent to the client
- which information should be presented and how
- introspection overhead should be minimal when the introspection is not used
- [consider reserving](https://github.com/eclipse-iceoryx/iceoryx/pull/1162#discussion_r813858455) the service string `iceoryx` for internal publishers only

## ToDo:

- [ ] Fix/Design or refactor error handling in `ProcessIntrospection::run()`

## Detailed information

When the introspection was implemented multiple pub/sub scenarios have not been considered, e.g. when a subscriber is subscribed to multiple publishers only one connection is displayed. When one publisher stops sending data then the subscriber is shown as disconnected (while still receiving data from the other publisher). This has to be adapted.
Additionally, introspection tests are missing and the existing ones may have to be adapted once it is clear how data is internally stored and sent to the client.

Open questions:

- Can we replace std::map?
- Do we want to have a mechanism to turn off the data transfer to the introspection? Do we still need it after the refactoring?
- Do we want to stick to ncurses?
- Also consider #414 and #31 and #402
- Which features that have been removed shall be re-added? See this [PR](https://github.com/eclipse-iceoryx/iceoryx/pull/360/files)
- Throughput calculation?
- Decide on whether we want parsable introspection output

Start by reading ProcessIntrospection::run() and the existing introspection tests, then map how data is stored and sent to the client across the listed pub/sub scenarios. Done means an agreed refactoring design, corrected multi-publisher behavior, minimal overhead when introspection is unused, and updated introspection tests; resolve the open questions before implementation.

Written by the indexing model from the issue text.