Add keep_alive() call when an error occurs in fetch_next(), to prevent permanent failure when redis restarts

[zakstucke]

As per title, without this change when redis restarts, it seems apalis get's stuck repeating this error over and over:

WARN  An error occurred during streaming jobs: An error was signalled by the server - ResponseError: user_script:15: consumer not registered script: e3ac73ae44f02d04176688b08c26ed751417cdb0, on @user_script:15.

After this change, on error keep_alive() will re-register, so it's fixed on the next call.

[geofmureithi]

I think this should only happen when one encounters a specific error, could you provide some logs or more context of what you are encountering?

[zakstucke]

You can reproduce by:

• Starting redis
• Starting apalis job queue
• send and process some jobs
• Kill redis
• Restart redis
• Send another job
• Apalis endlessly emits WARN An error occurred during streaming jobs: An error was signalled by the server - ResponseError: user_script:15: consumer not registered script: e3ac73ae44f02d04176688b08c26ed751417cdb0, on @user_script:15.

This error is coming from packages/apalis-redis/lua/get_jobs.lua line 15

By recalling keep_alive(), apalis can then recover and continue processing jobs, otherwise a single redis failure breaks apalis permanently

[zakstucke]

@geofmureithi updated