Fix inverted observer return-value javadoc in addNewSolutionObserver

[OscarSotoSanchez]

What

The method-level javadoc of BrkgaMpIpr.addNewSolutionObserver documented the observer's return value backwards. It said the observer returns true "if the algorithm should stop immediately", when the opposite is true.

Why it's wrong

The C++ core run loop (brkga_mp_ipr.hpp) drives the observer like this:

bool run = true;
while (run) {
    ...
    run &= callback(status);   // callback returns false -> run=false -> loop stops
}

So a callback returning false stops the optimization and true keeps it running. Verified empirically too (returning false cut the search right after the first improvement).

The corrected wording now matches every other (correct) place in the repo:

• NewSolutionObserver.onNewSolution @return — "true to keep running, false to stop"
• native bridge brkga_bridge.cpp — "non-zero to keep running (false stops the optimization)"
• the MainComplete examples (single & multi) — return true; // Don't stop the optimization.
• docs/GUIDE.md

Note

The inverted phrasing was inherited from the upstream C++ docs (codeurjc/brkga_mp_ipr_cpp), whose observer documentation contradicts its own run() loop in several spots. A separate issue there is worth filing.