Decouple headtracker tests

[dimriou]

Removed as many Core dependencies as we could from head tracker tests. Only package left is pgtest

The base branch was changed.

[jmank88]

Only package left is pgtest

Which part of the API is used? Some of that package can move to chainlink-common (or it may already have equivalents), but the txdb stuff is complex and worth investigating alternatives (which have already been exploring for core besides, without this extra complexity)

[dimriou]

Only package left is pgtest

Which part of the API is used? Some of that package can move to chainlink-common (or it may already have equivalents), but the txdb stuff is complex and worth investigating alternatives (which have already been exploring for core besides, without this extra complexity)

That's actually something I wanted to chat about at some point, but the biggest problem I see is that orm_test and similar files assume a db state that is currently located in Core. pgtest is used as a bootstraper for such tests so I'm not sure if there is an easy way to get around this even if we pull some code inside chainlink-common.
For example, in Head Tracker package, ther orm tests will try to execute the queries in a mock DB and they assume the state is already there but the db schema actually lives inside Core.

[jmank88]

That's actually something I wanted to chat about at some point, but the biggest problem I see is that orm_test and similar files assume a db state that is currently located in Core. pgtest is used as a bootstraper for such tests so I'm not sure if there is an easy way to get around this even if we pull some code inside chainlink-common. For example, in Head Tracker package, ther orm tests will try to execute the queries in a mock DB and they assume the state is already there but the db schema actually lives inside Core.

@krehermann might have some opinions