glideclientglobal_manifests generated too often in current configuration

[StevenCTimm]

glideclientglobal_manifests is being regenerated with every run of the resource_request channel every five minutes.
This is overkill, it only needs to be generated every time that the decision engine reconfigures, or that a secret changes.
It is difficult to debug in the current configuration because the block ends up being ephemeral and only there for a few seconds out of the 5-minute channel life time, making it very hard to dump and debug.

[StevenCTimm]

Bruno coimbra mentioned in glideinWMS meeting that there was a problem with the hashing function that the glideinwms frontend had been using in python3, because the hashing seed gets reset with every new process. Glideinwms has implemented a new hashing function. should see if this can be ported into the decision engine fork of this code.

[mambelli]

@StevenCTimm the hashing problem has been fixed in GlideinWMS. I see you reopened this. Can it be closed?

[StevenCTimm]

I closed it by accident on Jan 13.
The libs in question that do this hashing, inherit from glideinwms libs and we are now running 3.9.3
I haven't checked the underlying data blocks to see if the hashes change that often.

But it's more than a hashing issue.. it's the fact that we are generating glideclientglobal manifests on
every run of the resource_request channel when we really wouldn't need to.. they rarely change.