Fix VICIdial System Lagging and Heartbeat Issue

We have a self-hosted VICIdial system where agents are incorrectly becoming LAGGED/PAUSED because the agent heartbeat and last_update_time are not being refreshed reliably. We have already traced the issue significantly and do not want the troubleshooting restarted from scratch. The established chain is: Browser JavaScript refresh timer → AJAX request → conf_exten_check.php → PHP processing → agent heartbeat/refresh → last_update_time → VICIdial background processes → LAGGED status. We identified that the browser is repeatedly calling conf_exten_check.php, and the request returns HTTP 200 but an incorrect/blank response, which causes JavaScript processing to fail. We also found PHP warnings related to an undefined $DB variable in functions.php. Current investigation shows that vicidial.php mainly sets last_update_time when the vicidial_live_agents record is initially created, while the continuous heartbeat update mechanism still needs to be traced and repaired. Goal: identify the exact root cause of the failed agent heartbeat, restore reliable continuous last_update_time updates, ensure agents remain properly active/READY without falsely becoming LAGGED, and verify that autodialing works normally. The solution must be a permanent code/configuration fix, not a workaround such as manually updating MySQL timestamps, disabling LAGGED detection, or masking the symptom. The freelancer should be experienced with VICIdial, PHP, MySQL/MariaDB, JavaScript/AJAX, Asterisk and VICIdial background Perl processes and should document exactly what was changed and why.