dont sync metagraph on neuron init

[camfairchild]

This fixes errors that require the subtensor config to be passed on the CLI.

[ifrit98]

Great catch!

Will port this over to openminers also.

[p-ferreira]

LGTM although I'm not sure if I understood the implication on the validator.
So we want to sync the metagraph with self.subtensor?

The current implementation on openvalidators syncs like
self.metagraph.sync(). I see in the metagraph code that lite is set to true by default, but the subtensor is initially empty and later filled with bt.subtensor.

Should we replicate this approach to the openvalidators as well?

[ifrit98]

Should we replicate this approach to the openvalidators as well?

I believe we should, as (soon-ish) _neuron will be removed from bittensor proper with the next release following 5.2.1.

[ifrit98]

See #1426

[camfairchild]

LGTM although I'm not sure if I understood the implication on the validator.
So we want to sync the metagraph with self.subtensor?

The current implementation on openvalidators syncs like
self.metagraph.sync(). I see in the metagraph code that lite is set to true by default, but the subtensor is initially empty and later filled with bt.subtensor.

The config for the subtensor, including any chain endpoint, only gets passed to self.subtensor

We also need that when syncing the metagraph, so we have to pass that in somehow, meaning we should sync using it and disable the auto-sync when it initializes the object (because it has no config param)