fix(qe): spawn server in its own task (#5446)
There's something that has always been a problem in the QE binary, and
was missed in [#5444]: important long running work must not be done in
the future passed to `Runtime::block_on` (which includes the body of the
function marked as `#[tokio::main]`) when using a multi-threaded
runtime.
The reason for this is that this future is special: it is the only one
that is allowed to be `!Send` (which allows to hold `!Send` data in
`main`), but it runs directly on the main thread, and not in a worker
thread, and does not participate in the normal task scheduling and
work-stealing. Any code in there essentially becomes single-threaded,
except for the child tasks it spawns.
The [documentation][] says the following:
> Note that the async function marked with this macro does not run as a
> worker. The expectation is that other tasks are spawned by the
> function here. Awaiting on other futures from the function provided
> here will not perform as fast as those spawned as workers.
[#5444]: https://github.com/prisma/prisma-engines/pull/5444
[documentation]:
https://docs.rs/tokio-macros/latest/tokio_macros/attr.main.html#non-worker-async-function
