Change int to float for flDeadtime calculation

[xXxPROHvHeRxXx]

Description

This is my original work. I identified a bug in
CLagCompensationManager::FrameUpdatePostEntityThink() in
src/game/server/player_lagcompensation.cpp.

Bug

The variable flDeadtime is declared as int but stores the result
of a float subtraction:

int flDeadtime = gpGlobals->curtime - sv_maxunlag.GetFloat();

The fl prefix in the variable name suggests it was intended to be
a float. The implicit cast to int truncates the fractional part.

Impact

Lag compensation records are pruned too aggressively — up to 1 tick
earlier than intended.

Example:

• curtime = 150.7, sv_maxunlag = 1.0
• Expected (float): flDeadtime = 149.7
• Actual (int): flDeadtime = 149

Records with m_flSimulationTime between 149.0 and 149.7 are
incorrectly removed. This can reduce hit registration accuracy
for players at higher latencies.

Fix

- int flDeadtime = gpGlobals->curtime - sv_maxunlag.GetFloat();

+ float flDeadtime = gpGlobals->curtime - sv_maxunlag.GetFloat();

Confirmation

This Contribution is my original work. I identified this issue
by reviewing the publicly available Source SDK 2013 source code
in this repository.