Abstract environment layer to allow different environments per project client / project details

Early patch.

Updated

Fully implemented interface and API.

Initial commit of #3.

I don't want to have to block when an environment is currently testing a patch. As such the system needs to keep track of what clients are testing a patch so that it can be sent to as many as possible without blocking.

I'm thinking something simple (only a record when current testing):

environment_status
  test_id: test ID related to status information
  environment_id: environment ID related to status information
  client_id: client ID that test was sent to

The status of each environment can be displayed as another fieldset on the test page.

Once a client requests a test a record is inserted into environment_status, once a result is returned the record is removed.

The query to find eligible tests should ignore them if a result or environment_status record exists.

For simplicity, query all tests that have not already been tested in the client's environment and are "queued". Loop through them and get the relevant environments. If the client is in the list then return the test.

The current test code:

define('PIFR_SERVER_TEST_STATUS_STUB', 1);
define('PIFR_SERVER_TEST_STATUS_QUEUED', 2);
define('PIFR_SERVER_TEST_STATUS_POSTPONED', 3);
define('PIFR_SERVER_TEST_STATUS_SENT', 4);
define('PIFR_SERVER_TEST_STATUS_RESULT', 5);

will need to be changed since there will no longer be a need for the SENT status. Tests will always be QUEUED until they have been tested on all relevant environments or received a fail on a required environment.

Also remove the client_id of the pift_test table since that will be recorded in environment_status.

Much further integration and such.

Further abstraction.

Appears to be pretty much there.

The majority of things seems to be working....haven't got through everything, but this is complete enough to commit.

Committed.