[19743] Add expiration checks in database fragile pointers

[Carlosespicur]

Main changes

This PR fixes a crash reported in Fast DDS Monitor that can be reproduced by following the steps below:

1. Run an endpoint and open Fast DDS Monitor
2. Select a locator of the previous endpoint
3. Delete the endpoint or locator (it will become inactive in the monitor)
4. Delete inactive entities

In the database, associated locator entities are stored as fragile references (fragile_ptr) but returned as shared references (shared_ptr). When an associated locator entity (in this case a subscriber or endpoint) is deleted, its associated fragile_ptr becomes expired and is returned as nullptr when casted to shared_ptr, causing memory errors when trying to access an object member.

There are two ways to deal with this problem:

1. Check whether fragile_ptr is expired or not before casting to shared_ptr.
2. Avoid using shared pointers and use fragile_ptrs instead. fragile_ptr is designed to avoid these memory errors without breaking the API.

The base branch was changed.