Setting E does not mod by 2pi, resulting in convergence errors

[mgiuca]

Steps to reproduce:

Run the following program:

from numpy import degrees, radians
from orbital import earth, KeplerianElements

orbit = KeplerianElements(a=1000.0, e=0.75, body=earth)

orbit.E = radians(485.140095)
print('orbit.M = %f°' % degrees(orbit.M))
print('orbit.E = %f°' % degrees(orbit.E))

Expected behaviour:

(Ignoring the OrbitalWarning that is always printed when using the E setter.)

orbit.M = 90.000000°
orbit.E = 125.140095°

Actual behaviour:

orbit.M = 450.000000°
Traceback (most recent call last):
  File "set-e-error.py", line 8, in <module>
    print('orbit.E = %f°' % degrees(orbit.E))
                                    ^^^^^^^
  File "orbital/elements.py", line 200, in E
    return eccentric_anomaly_from_mean(self.e, self._M)
           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "orbital/utilities.py", line 99, in eccentric_anomaly_from_mean
    raise ConvergenceError('Did not converge after {n} iterations. (e={e!r}, M={M!r})'.format(n=MAX_ITERATIONS, e=e, M=M))
orbital.utilities.ConvergenceError: Did not converge after 100 iterations. (e=0.75, M=np.float64(7.853981640431436))

This is because the E setter (or utilities.mean_anomaly_from_eccentric) does not mod the resulting value of M by 2pi, so M is set to 450° instead of 90°, and then the E getter has unexpected behaviour when using the value of M outside the range. Note that the convergence error only occurs for sufficiently eccentric orbits (it does not crash when e=0.0).

Note that both the M and f setters do mod by 2pi, it is only the E setter that fails to do so.