`<numeric>`: error C3198: invalid use of floating-point pragmas: fenv_access pragma operates only in precise mode

Describe the bug
Enabling fenv_access prior to including <numeric> triggers error C3198.

Command-line test case

// float_control.cpp
#pragma fenv_access(on)
#include <numeric>

int main() {}

D:\>cl /std:c++17 float_control.cpp
Microsoft (R) C/C++ Optimizing Compiler Version 19.31.31104 for x64
Copyright (C) Microsoft Corporation.  All rights reserved.

float_control.cpp
C:\Program Files\Microsoft Visual Studio\2022\Community\VC\Tools\MSVC\14.31.31103\include\numeric(52): error C3198: invalid use of floating-point pragmas: fenv_access pragma operates only in precise mode
C:\Program Files\Microsoft Visual Studio\2022\Community\VC\Tools\MSVC\14.31.31103\include\numeric(158): error C3198: invalid use of floating-point pragmas: fenv_access pragma operates only in precise mode

Expected behavior
No error.

STL version

Microsoft Visual Studio Community 2022 (64-bit) - Current
Version 17.1.1

Start with the provided float_control.cpp reproduction and inspect the header locations reported at lines 52 and 158. Run the MSVC command shown with /std:c++17, then trace the floating-point pragma handling in those header sections. Done means enabling fenv_access before including no longer produces C3198.