Modification based dataset realtime update

[kangz12345]

I apologize for the large diffs... It was an urgent patch so I didn't break it to several PRs.

To summarize this PR:

1. Change polling strategy: periodic polling -> long polling
2. Change dataset update strategy: fetch whole dataset every time -> only modifications since the last fetch
3. Change threading strategy: Spawn a new thread for every fetch -> Spawn a thread for each sync (spawn once start sync, destroy after stop sync) and run a long polling loop
4. Synchronize fetching thread and GUI thread with a QMutex and a QWaitCondition
5. NOT changed the data structure or plot strategy: it still updates the plot for the whole dataset when the dataset is updated (even by modifications)

More description for the thread synchronization:

• Without any synchronization, the fetching thread will report new modifications continuously while the GUI suffers from organizing and plotting the dataset.
• As the dataset size grows, the GUI latency becomes longer and the modified signals start to be queued.
• Then the GUI event queue will be full of modified signals, so the GUI might freeze or glitches may appear.
• Therefore, using a mutex and wait condition, the fetcher thread should wait for the GUI thread to update the plot before fetching the next modifications.

As I tested, the most of the latency was because of the large data transfer. Therefore, I didn't change the dataset organizing and plotting policy.

This branch is already tested and used in the experiment PC. It works well without any glitch up to 200,000 data points (tested so far, will work for even more!) (the previous version crashed for 20,000 data points).

This branch works after snu-quiqcl/artiq-proxy#93 is done (already done in a not-merged branch).

This closes #207.

[kangz12345]

There comes a little bit of change soon, because of a bug found last week. It is resolved in the lab PC, so I will apply it to this branch soon.

[kangz12345]

There comes a little bit of change soon, because of a bug found last week. It is resolved in the lab PC, so I will apply it to this branch soon.

I will create another PR for the updates.