gh-NNNN: Remove _Py_dg_strtod() from dtoa.c

_Py_dg_strtod() and its helper functions (s2b, hi0bits used only by
s2b, b2d, sd2b, ratio, BCinfo, sulp, bigcomp, plus the tens/bigtens
tables used only by strtod) are removed — 1075 net lines deleted.

String-to-double conversion is now handled by _PyWuffs_strtod() (added
in the previous commit) which combines the Eisel-Lemire fast path with
Wuffs's High Precision Decimal fallback for 100% correct rounding.

The Bigint infrastructure (Balloc, Bfree, multadd, mult, lshift, diff,
cmp, pow5mult, i2b) and the functions still needed by _Py_dg_dtoa
(hi0bits, lo0bits, d2b, tens, bigtens, dshift, quorem) are retained.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

Author: eendebakpt

Include/internal/pycore_dtoa.h

@@ -24,7 +24,6 @@ extern "C" {
     }
 #endif
 
-extern double _Py_dg_strtod(const char *str, char **ptr);
 extern char* _Py_dg_dtoa(double d, int mode, int ndigits,
                          int *decpt, int *sign, char **rve);
 extern void _Py_dg_freedtoa(char *s);

Include/internal/pycore_pymath.h

@@ -59,14 +59,15 @@ static inline void _Py_ADJUST_ERANGE2(double x, double y)
 
 //--- HAVE_PY_SET_53BIT_PRECISION macro ------------------------------------
 //
-// The functions _Py_dg_strtod() and _Py_dg_dtoa() in Python/dtoa.c (which are
-// required to support the short float repr introduced in Python 3.1) require
-// that the floating-point unit that's being used for arithmetic operations on
-// C doubles is set to use 53-bit precision.  It also requires that the FPU
-// rounding mode is round-half-to-even, but that's less often an issue.
+// The function _Py_dg_dtoa() in Python/dtoa.c (which is required to support
+// the short float repr introduced in Python 3.1) and _PyWuffs_strtod() in
+// Python/_ryu/pystrtod_wuffs.h require that the floating-point unit that's
+// being used for arithmetic operations on C doubles is set to use 53-bit
+// precision.  It also requires that the FPU rounding mode is
+// round-half-to-even, but that's less often an issue.
 //
 // If your FPU isn't already set to 53-bit precision/round-half-to-even, and
-// you want to make use of _Py_dg_strtod() and _Py_dg_dtoa(), then you should:
+// you want to make use of _PyWuffs_strtod() and _Py_dg_dtoa(), then you should:
 //
 //     #define HAVE_PY_SET_53BIT_PRECISION 1
 //