[mccall_model] Replace np.sum(a * b) with a @ b

[UdohNakamura]

Replace np.sum(a * b) with a @ b
#463

[mmcky]

thank you @UdohNakamura for this FIX - I have enabled the tests to run.

[HumphreyYang]

Many thanks @UdohNakamura and @mmcky,

When building on my side I got

<string>:3: NumbaPerformanceWarning: '@' is faster on contiguous arrays, called on (Array(float64, 1, 'C', False, aligned=True), Array(float64, 1, 'A', False, aligned=True))

Upon checking, I found we annotate the type

    ('w', float64[:]),   # array of wage values, w[i] = wage at state i
    ('q', float64[:])    # array of probabilities

The following type will help (see more here:

    ('w', float64[::1]),   # array of wage values, w[i] = wage at state i
    ('q', float64[::1])    # array of probabilities

Please let me know what you think.

[oyamad]

@HumphreyYang Thanks, I didn't know that. Maybe write float64[::1] in place of float64[:] and explain that it is needed to have @ work inside a jitclass?