Get rid CL_REQUIRE/CL_PROVIDE(cl_F_leastneg), it is not really necessary.

Move static variables least_negative_[SFDL]F into least_negative_LF() function
(which is the only user of those variables) in order to avoid possible static
order initialization problems. While at it, make least_negative_LF() function
static.

include/cln/float.h

@@ -567,7 +567,6 @@ extern const cl_F least_positive_float (float_format_t f);
 // Returns the least negative floating point number (i.e. < 0 but closest to 0)
 // in float format f.
 extern const cl_F least_negative_float (float_format_t f);
-//CL_REQUIRE(cl_F_leastneg)
 
 // Returns the smallest floating point number e > 0 such that 1+e != 1.
 extern const cl_F float_epsilon (float_format_t f);

src/float/misc/cl_F_leastneg.cc

@@ -3,8 +3,6 @@
 // General includes.
 #include "cl_sysdep.h"
 
-CL_PROVIDE(cl_F_leastneg)
-
 // Specification.
 #include "cln/float.h"
 
@@ -19,6 +17,21 @@ CL_PROVIDE(cl_F_leastneg)
 
 namespace cln {
 
+static inline const cl_LF least_negative_LF (uintC len)
+{
+	var Lfloat erg = allocate_lfloat(len,LF_exp_low,-1);
+	#if CL_DS_BIG_ENDIAN_P
+	  TheLfloat(erg)->data[0] = bit(intDsize-1);
+	  clear_loop_up(&TheLfloat(erg)->data[1],len-1);
+	#else
+	  var uintD* ptr = clear_loop_up(&TheLfloat(erg)->data[0],len-1);
+	  *ptr = bit(intDsize-1);
+	#endif
+	return erg;
+}
+
+const cl_F least_negative_float (float_format_t f)
+{
 	// Exponent so klein wie möglich, Mantisse 10...0, Vorzeichen -.
 
 	static const cl_SF least_negative_SF =
@@ -34,21 +47,6 @@ static const cl_DF least_negative_DF =
 		encode_DF(-1,DF_exp_low-DF_exp_mid,bit(DF_mant_len-32),0);
 	#endif
 
-inline const cl_LF least_negative_LF (uintC len)
-{
-	var Lfloat erg = allocate_lfloat(len,LF_exp_low,-1);
-	#if CL_DS_BIG_ENDIAN_P
-	  TheLfloat(erg)->data[0] = bit(intDsize-1);
-	  clear_loop_up(&TheLfloat(erg)->data[1],len-1);
-	#else
-	  var uintD* ptr = clear_loop_up(&TheLfloat(erg)->data[0],len-1);
-	  *ptr = bit(intDsize-1);
-	#endif
-	return erg;
-}
-
-const cl_F least_negative_float (float_format_t f)
-{
 	floatformatcase((uintC)f
 	,	return least_negative_SF;
 	,	return least_negative_FF;
@@ -59,4 +57,3 @@ const cl_F least_negative_float (float_format_t f)
 
 }  // namespace cln
 
-CL_PROVIDE_END(cl_F_leastneg)