Fix race conditions in UseIModelData

[simihartstein]

The iModelGrid component was erroneously saying that I had no iModels in my iTwin. This was happening roughly 90% of the time with my initial code, but an unrelated code change could cause it to happen <5% of the time.

The race condition came in due to the useIModelData hook having multiple useEffects that relied on the same dependencies, changed the same state, where the order mattered. React does not guarantee an order in which these effects are run.

The component was refactored and simplified to have more linear logic without race conditions.

Other bugs fixed in the process of testing my changes:

• iModels were only being fetched when the visibility of the ghost tile changed. If the query didn't result in the visibility changing (an unlikely occurrence with a page size of 100), then the iModelGrid would be stuck.
• MaxCount was calculated incorrectly while not on the first page.

[simihartstein]

I have not tested in React 18 strict mode, but there is a chance this PR could solve #128, as that looks like it could be a race condition.

The abort controller was prematurely terminated when the dependencies of the effect were changed before the request finished.

[simihartstein]

I found some bugs when the limit is set low and the imodels don't fill the page. I presume it's related to my changes even though my changes were in the data layer. Looking into it...

[simihartstein]

Fixed a couple more bugs.

• iModels were only being fetched when the visibility of the ghost tile changed. If the query didn't result in the visibility changing (an unlikely occurrence with a page size of 100), then the iModelGrid would be stuck.
• MaxCount was calculated incorrectly while not on the first page.

[aruniverse]

didnt test locally but seems ok

[aruniverse]

didnt test locally but seems ok