Unexpected return from hyp1f1 when evaluating limit at negative infinity

[jacksonvanover]

Though mpmath appears to support evaluating limits approaching positive infinity, matching the results returned by SciPy and GSL, and jmat, it diverges from those three libraries when evaluating the limit at negative infinity with the first argument as 1; mp.hyp1f1 returns NaN when the other libraries agree on zero.

>>> special.hyp1f1(0, 1, np.inf)
1.0
>>> mp.hyp1f1(0, 1, np.inf)
mpf('1.0')
>>> special.hyp1f1(0, 1, np.NINF)
1.0
>>> mp.hyp1f1(0, 1, np.NINF)
mpf('1.0')
>>> special.hyp1f1(1, 2, np.NINF)
0.0
>>> mp.hyp1f1(1, 2, np.NINF)
mpf('nan')
>>> special.hyp1f1(2, 2, np.NINF)
0.0
>>> mp.hyp1f1(2, 2, np.NINF)
mpf('0.0')
>>> mp.hyp1f1(1, 5, np.NINF)
mpf('nan')
>>>