Fix possible crash with trail systems

[slipher]

I received a couple of a user's crash dumps on 0.52.1 looking like this:

Thread 0 (crashed)
 0  main.nexe!CG_DrawActiveFrame(int, bool) [cg_attachment.cpp : 305 + 0x2]
    rax = 0x0000000000000000   rdx = 0x0000000000000003
    rcx = 0x0000000000000000   rbx = 0x0000000000000000
    rsi = 0x000000001116a814   rdi = 0x0000000011155afc
    rbp = 0x00000000fffeff80   rsp = 0x00000000fffdf8c0
     r8 = 0x0000000000000000    r9 = 0x0000000000000128
    r10 = 0x00000000fcbd6000   r11 = 0x00000466000d7680
    r12 = 0x0000000000000003   r13 = 0x0000000000000000
    r14 = 0x00000000000cd047   r15 = 0x0000046600000000
    rip = 0x00000000000d7799
    Found by: given as instruction pointer in context
 1  main.nexe!VM::VMHandleSyscall(unsigned int, Util::Reader) [cg_api.cpp : 92 + 0x20]
    rbx = 0x0000000000000000   rbp = 0x00000000fffeff80
    rsp = 0x00000000fffe2ff0   r12 = 0x00000000fffefeb0
    r13 = 0x0000000000000064   r14 = 0x0000000000000000
    r15 = 0x0000046600000000   rip = 0x0000000000028f20
    Found by: call frame info
 2  main.nexe!main [VMMain.cpp : 66 + 0x20]
    rbx = 0x00000000fcb53f78   rbp = 0x00000000fffeff80
    rsp = 0x00000000fffefe70   r12 = 0x00000000fffefeb0
    r13 = 0x0000000000000064   r14 = 0x00000000fffefe90
    r15 = 0x0000046600000000   rip = 0x00000000001ca2e0
    Found by: call frame info
 3  main.nexe!_start [start.c : 68 + 0x39]
    rbx = 0x0000000000000015   rbp = 0x00000000fffeff80
    rsp = 0x00000000fffeff30   r12 = 0x00000000fffeffbc
    r13 = 0x00000000fffeffc8   r14 = 0x0000000000000015
    r15 = 0x0000046600000000   rip = 0x00000000004aebc0
    Found by: call frame info

I suspect it was caused by this part of my recent changes to CG_DestroyTrailSystem:

-       ts = nullptr;
+       *ts = nullptr;

The change is fixing an obvious typo, but it turns out making the code behave according to the author's intention here was a bad idea.

Still no idea why it consistently happens to some people, but not at all to others.

[slipher]

lautre on IRC has discovered a way to reproduce the bug: select the "Lowest" graphics preset. ("Low" or higher prevents it.) Thanks to this tip I was able to confirm that the crash happens in CG_UpdateBeam as I suspected, and that this PR fixes it.

[illwieckz]

That code is now running on both unvanquished.net and gg.illwieckz.net servers since multiple days and it looks like it fixes the bug for everyone affected.