Improve random seed setting for all examples

[muupan]

in the same way as #239

• ALE accepts [0,2**31), so ALE examples use [0,2**31).
• Although gym envs accepts broader range of values, numpy.random.seed accepts [0,2**32) as a integer, so gym examples use [0,2**32).
• Logger outputs of some examples were invisible, so I changed their logger settings.

[toslunar]

LGTM