Do not apply browser timeout option as default navigation timeout

[ka3de]

Currently the timeout option defined at the browser level applies to two different functionalities:

• Sets the TO to apply for some "browser initialization" actions, such as TO when launching the browser instance.
• Sets the default navigation TO. This can be overwritten at the browserContext and page levels.

This double purpose is wrong, as a the use cases are quite different and a single value for both does not make sense. At the same time, though, we still believe that defining a TO for the browser initialization actions provides value and is required. Taking this into account our decisions have been:

• The browser level timeout option will apply only for the browser initialization actions. Not for the default navigation TO.
• The browser level timeout option will be moved to be defined as an ENV variable, instead of being defined at the script level.
• The default navigation TO will be set by our own implementation with an initial sensible value. This value will be able to be overwritten at the browserContext and page levels, as it's currently possible.

Related: #856.

[ka3de]

There was a misunderstanding in the analysis that resulted in this issue based on the documentation and the bug described in #865 , which, considering that most tests start with a navigation, behaves in a misleading manner which makes it look like it's the navigation TO the one that makes the test fail, but instead is the incorrectly set browser process TO.

Currently the browser.timeout is not set as the default navigation TO, instead, the default navigation TO applied to page and frame is inherited from browserContext's timeoutSettings, which are initialized to nil in its constructor, making it effectively use the default timeout of 30s.

Therefore, no exact fix is required for this, and instead fixing issue #865 will achieve the desired behavior.