fix: prevent crash on malformed FQDNs by handling LocationParseError (fixes #2970)

[marcos-dataops]

Description

This PR fixes the fatal crash reported in #2970 when searching for usernames ending in a period (e.g., alice.).

The Root Cause:
When a username ending in a period is interpolated into a subdomain-based target (like https://{}.empretienda.com.ar), it generates a malformed FQDN containing consecutive dots (alice..empretienda.com.ar). When requests_futures attempts to resolve this, the underlying urllib3 library raises a LocationParseError. Because this specific exception does not inherit from requests.exceptions.RequestException, it bypassed the existing error handling in get_response, crashing the entire application.

The Solution:
Instead of playing whack-a-mole by adding regex filters to individual targets in data.json (which breaks CI target validation as seen in PR #2971 and doesn't scale for future target additions), this PR implements a robust architectural fix at the network layer.

By catching urllib3.exceptions.LocationParseError and requests.exceptions.InvalidURL directly in the asynchronous future resolution block (sherlock_project/sherlock.py), the application now gracefully handles any malformed URL generation, marks the specific target as failed ("Illegal URL"), and continues execution without crashing.

Steps to Test

1. Run python3 -m sherlock_project alice.
2. Prior to this PR, the application would fatal crash immediately on alice..empretienda.com.ar.
3. With this PR, the error is caught, handled gracefully, and the scan finishes successfully.

Checklist:

• I have tested this code.
• I have updated the documentation (if necessary).
• This is not a target addition/modification (no data.json changes).