feature: set IN_JUMP for m_afButtonPressed in CBasePlayer::Jump to fix the incorrect behavior of bxt_autojump

[SmileyAG]

Current behaviour with bxt_autojump:

• You don't get extra speed from jumping on a func_conveyor entity (https://github.com/ValveSoftware/halflife/blob/c7240b965743a53a29491dd49320c88eecf6257b/dlls/player.cpp#L1643-L1648)
• Jump animation does not play in third person (https://github.com/ValveSoftware/halflife/blob/c7240b965743a53a29491dd49320c88eecf6257b/dlls/player.cpp#L1633-L1641)
• In other modifications, such as Cry of Fear, this has a more disastrous consequence (e.g. stamina is not spent when jumping)

This happens because of this check in the CBasePlayer::Jump code: https://github.com/ValveSoftware/halflife/blob/c7240b965743a53a29491dd49320c88eecf6257b/dlls/player.cpp#L1619-L1621

Proposed fix:

    int *m_afButtonPressed;
    bool is_buttonpressed_jump_set = false;

    if (offm_afButtonPressed)
    {
        m_afButtonPressed = reinterpret_cast<int*>(reinterpret_cast<uintptr_t>(thisptr) + offm_afButtonPressed);
        if (!(*m_afButtonPressed & IN_JUMP))
        {
            if (CVars::bxt_autojump.GetBool())
            {
              *m_afButtonPressed |= IN_JUMP;
              is_buttonpressed_jump_set = true;
            }
        }
    }

    insideCBasePlayerJump = true;
    ORIG_CBasePlayer__Jump(thisptr);
    insideCBasePlayerJump = false;
    
    if (is_buttonpressed_jump_set)
        *m_afButtonPressed &= ~IN_JUMP;

At the same time, this will also add the ability to use bxt_autojump in pre-HLSDK 2.0 game versions (where the movement code was accessed only for server-side)

I had previously done byte patching for CoF to fix that behavior in BXT code:

BunnymodXT/BunnymodXT/modules/ServerDLL.cpp

Lines 1832 to 1836 in 1196334

	} else { // CoF-Mod
	if ((*reinterpret_cast<byte*>(pCBasePlayer__Jump_OldButtons_Check_Byte) == 0x75) && CVars::bxt_autojump.GetBool())
	MemUtils::ReplaceBytes(reinterpret_cast<void*>(pCBasePlayer__Jump_OldButtons_Check_Byte), 1, reinterpret_cast<const byte*>("\xEB"));
	else if ((*reinterpret_cast<byte*>(pCBasePlayer__Jump_OldButtons_Check_Byte) == 0xEB) && !CVars::bxt_autojump.GetBool())
	MemUtils::ReplaceBytes(reinterpret_cast<void*>(pCBasePlayer__Jump_OldButtons_Check_Byte), 1, reinterpret_cast<const byte*>("\x75"));

But looking back I understand that it was a completely bad decision to fix it in that way, so that byte patching code can be removed as soon as the m_afButtonPressed set solution would be added