I've seen reports of the Fitbit HR being off by 25 BPM. I'm not familiar with the Dr. Coca pulse test, but the paper you link to mentions pulse rates that differ by only 10 BPM. It appears that consumer Photoplethysmogram monitors work best when the patient is inactive. I suspect the accuracy also depends on your physiology. I would recommend trying a wrist monitor such as the Fitbit HR, a chest strap monitor, and taking your pulse manually. Do this while you're inactive and compare the results. If the Fitbit HR is off from your manual measurement by 25 BPM, it sounds like it won't be useful for the Dr. Coca pulse test.
There are other options for monitoring heart rate that are very accurate:
- A holter monitor. These are usually prescribed devices--the hospital or physician loans them to you (though it appears you can get them on eBay).
- An Implantable Cardiac Monitor (ICM) (also known as an Implantable Loop Recorder (ILR)). This is an implantable device similar to a pacemaker, however it only measures your heart activity--it doesn't provide therapy (electrical stimulus). Some examples are the Biotronik BioMonitor and the Medtronik Reveal LINQ.
- There are disposable holter monitor patches that are coming onto the market. Many of these are still in clinical trials, however. The advantage of these devices is that they don't have to be implanted and you don't have to wear a bulky unit on your belt.
All three of the options above are used for diagnosing heart conditions by looking at ECG signals--they're probably overkill for your application, but they would work.