Handle timeout in storage properly

[lars-petter-hauge]

In the scenario of large files as records in storage, the default timeout of 5 seconds is not enough. In https://github.com/equinor/ert/pull/2729 this was increased to the magical number of 60. However handling timeout should be addressed in a better manner.

[lars-petter-hauge]

The current implementation is very rough, simply passing the same timeout to any type of connection issues to the requests (and having it as a total for aggregated time over an entire query). It is possible to have granular control, specifying timeout for connection, read (by chunk), write etc independently Fine tune configuration

We can set this at a much higher level as well, on the client initialisation for instance