Adjusted workes to threads logic to agree with what is in scipy.fft

[oleksandr-pavlyk]

@peterbell10

In [1]: import numpy as np, mkl_fft, mkl_fft._scipy_fft_backend as mkl_be, scipy, scipy.fft, mkl

In [2]: x = np.random.randn(8*7).reshape((7, 8))
   ...: with scipy.fft.set_backend(mkl_be, only=True):
   ...:     ff = scipy.fft.fft2(x, workers=-4)
   ...: ff2 = scipy.fft.fft2(x)
MKL_VERBOSE Intel(R) MKL 2020.0 Product build 20191102 for Intel(R) 64 architecture Intel(R) Advanced Vector Extensions 2 (Intel(R) AVX2) enabled processors
, Lnx 2.40GHz intel_thread
MKL_VERBOSE FFT(drfo7:8:8x8:1:1,bScale:0.0178571,tLim:1,desc:0x56202b943cc0) 37.40us CNR:OFF Dyn:1 FastMM:1 TID:0  NThr:16,FFT:13

[oleksandr-pavlyk]

Negative worker value is now interpreted relative to os.cpu_count() value, but if it exceeds Intel(R) MKL's own get_max_threads(), a one-time warning is issued:

(t_scipy-1.4.0) [14:40:32 linbox mkl_fft]$ ipython
Python 3.7.5 (default, Nov 23 2019, 04:02:01)
Type 'copyright', 'credits' or 'license' for more information
IPython 7.11.1 -- An enhanced Interactive Python. Type '?' for help.

In [1]: import numpy as np, mkl_fft, mkl_fft._scipy_fft_backend as mkl_be, scipy, scipy.fft, mkl

In [2]: x = np.random.randn(8*7).reshape((7, 8))
   ...: with scipy.fft.set_backend(mkl_be, only=True):
   ...:     ff = scipy.fft.fft2(x, workers=-4)
   ...: ff2 = scipy.fft.fft2(x)
/tmp/mkl_fft/_scipy_fft_backend.py:59: UserWarning: os.cpu_count() returned value of 32 greater than mkl.get_max_threads()'s value of 16. Using negative values of worker option may amount to requesting more threads than Intel(R) MKL can acommodate.
  ).format(self.cpu_count, max_threads))

In [3]: x = np.random.randn(8*7).reshape((7, 8))
   ...: with scipy.fft.set_backend(mkl_be, only=True):
   ...:     ff = scipy.fft.fft2(x, workers=-4)
   ...: ff2 = scipy.fft.fft2(x)

In [4]: np.allclose(ff, ff2)
Out[4]: True