SCNRM2 and DZNRM2 fail on master

[grisuthedragon]

Description
The SCNRM2 and the DZNRM2 fail the tests some when after the 3.12 release, broken in the current master.

 Test of subprogram number  6            SCNRM2
                                       FAIL
 SCNRM2: N=     1, INCX=  -2, IV= 9, IW= 1, test=            NaN
 SCNRM2: N=     1, INCX=  -2, IV= 9, IW= 2, test=            NaN
 SCNRM2: N=     1, INCX=  -2, IV= 9, IW= 3, test=            NaN
 SCNRM2: N=     1, INCX=  -2, IV= 9, IW= 4, test=            NaN
 SCNRM2: N=     1, INCX=  -2, IV= 9, IW= 5, test=            NaN
 SCNRM2: N=     1, INCX=  -2, IV= 9, IW= 6, test=            NaN
 SCNRM2: N=     1, INCX=  -2, IV= 9, IW= 7, test=            NaN
 SCNRM2: N=     1, INCX=  -2, IV= 9, IW= 8, test=            NaN
 SCNRM2: N=     1, INCX=  -2, IV= 9, IW= 9, test=            NaN
 SCNRM2: N=     1, INCX=  -2, IV= 9, IW=10, test=            NaN
                                       FAIL
...

The output for DZNRM2 is similar.

Tested with

• Ubuntu 20.04, gcc/gfortran 9
• Ubuntu 20.04, gcc/gfortran 10
• Ubuntu 24.04, gcc/gfortran 13

Going back to the 3.12.0 everything works fine.

[martin-frbg]

Curious as the only remotely related change seems to be #964 (which modifies some of the added test code for ?NRM2 from #925 that went in a week or two before the release).