feature: `get_column_names` accepts a dtypes argument

[JovanVeljanoski]

Now we can query column names by data type:

import vaex

df = vaex.datasets.titanic()
df.get_column_names(dtypes=[float])

# returns  ['age', 'fare', 'body']

Several attempts at this have been done at the past. This is my latest proposal and I think it works quite well. Closes:

• https://github.com/vaexio/vaex/issues/240
• https://github.com/vaexio/vaex/pull/251