Fix OOM due to update avalanches

[RealDolos]

Problem

• The proxy object checks if Date.now() - proxyObject.lastUpdateCheck > updateTimer on every property access.
• lastUpdateCheck starts out as 0 and is only updated on receiving a checking event.
• The checking event is delayed by at least 200 ms in .checkUpdates() (running after_wait(200))
• Therefore the above check holds true at least for the first 200ms of the process (or a little more, until the checking event is actually done emitting)
• Therefore causing an avalanche of checkUpdates() running in "parallel", once for every property access on that proxied object while lastUpdateCheck is still 0 during those first 200ms.

Result

Process went up to consume 20GB of physical memory, swapping out most the physical memory of most other processes and eventually getting OOM killed by the kernel. Oops.

Fix

Remove the lastUpdateCheck/checkUpdates() calls from the proxy object completely. The UpdateSubscriber will perform these anyway already. No need to have this triggered on property access, kinda redundant.

Also, move out the proxyObject.subscriber setup from the proxy get into the main wrapReader body. No need to permanently check the creation when it's created exactly once anyway.

Also, triggerUpdate() is useless; stubbed.

[GitSquared]

Thanks for the PR. Will release 1.0.7 in a few minutes. :+1:

[RealDolos]

Update tests

I knew I wanted to do something more before submitting the PR o_O

[GitSquared]

No problem, tests are a pain in the *** to update, I don't blame you :smile: