`<stacktrace>`: stacktraces deeper than `USHORT_MAX` are truncated to `USHORT_MAX` by `stacktrace::current()`

Describe the bug
see title - stacktraces grabbed by stacktrace::current() will always be at most the most recent USHORT_MAX entries.

This requires some amount of investigation to solve:

• Is this a problem seen by real code? Do we even really need to worry about this?
• Is this something we can solve? Can we loop over CaptureStackBackTrace() with Skip += 0xFFFF each time?
  • if not, should we fail in this case?

Command-line test case

C:\Temp>type stacktrace-deep.cpp
#include <stacktrace>
#include <iostream>

void print_stacktrace()
{
    auto s = std::stacktrace::current();
    if (s.empty())
    {
        std::cout << "failed to initialize\n";
    }
    else
    {
        std::cout << *s.rbegin() << '\n';
    }
}

#pragma optimize("", off) // inhibit TCO
void stackin(unsigned current_frame)
{
    if (current_frame == 0x10000)
    {
        print_stacktrace();
    }
    else
    {
        stackin(current_frame + 1);
    }
}
#pragma optimize("", on)

int main()
{
    stackin(0);
}

C:\Temp> cl /EHsc /W4 /WX /std:c++latest .\stacktrace-deep.cpp /F0x400000 /Z7 /nologo

stacktrace-deep.cpp
C:\Temp>.\stacktrace-deep.exe
C:\Temp\stacktrace-deep.cpp(26): stacktrace_deep!stackin+0x30

Expected behavior
It should print ntdll!RtlUserThreadStart+0x48

Start by inspecting the implementation of stacktrace::current() and its use of CaptureStackBackTrace(). Run the provided stacktrace-deep.cpp command-line reproduction with a 0x10000-frame recursion depth, then determine how deep frames should be handled. Done means the test reaches and reports ntdll!RtlUserThreadStart+0x48 instead of truncating at stackin.