Simple answer. Yes, each researcher keeps data they use for research for at least 5 years, but their is no central database. You will need to request access to that data from the authors of any academic research on lie detection. If you read what I have written below then you can follow the links to the authors of the research and request their data. Although please be aware that they have may not wish to share their data.
The kinds of data you ask do exist but nobody takes all these measures in one experiment.
I feel inclined to provide a bit of background about lie detection research if you might indulge me.
Firstly, there is no such thing as a lie detector. The most sensitive equipment that could potentially be used to detect lies is electroencephalography (EEG), as it has high temporal resolution which would be necessary for lie detection. EEG equipment records electrical signals on the the scalp, which originate from the brain if you do it well, or from the surrounding muscles if you set the equipment up badly. Within the last 20 years research has tried and failed to accurately and constantly detect detection with both EEG and functional magnetic resonance imaging (fMRI). Hippocampal activity you will only really get with fMRI, although there are methods for EEG to get a weak signal. In general there are many problems using fMRI for lie detection, please see this link. Both methods are inconsistent at the detection of deception at the level of the individual, although some methods claim to have more success than others(please bare in mind this research was done in an Iranian university, meaning they are not known for their academic rigour at the time of writing). For instance this attached review reports that EEG analysis detection of 'deception' within p300 (this is a decision making signal that EEG detects) has been reported between 90 and 75% accurate in participant groups, looking at the data I would put it closer to 70%. This is not good enough for any uses by law enforcement.
With regards to questioning, law enforcement uses very specific techniques i.e. the cognitive interview so as not to negatively impact on recall. In addition these experiments are often done in the lab, where students are given a scenario and told to lie. This is very different from lying in the real world, for instance the signal of 'deception' in the p300 may get easily confused with normal decision making. Or the activity in certain neural areas may be normal functioning involved in recall rather than deception.
Without training lie detection is around 50%, in laboratory conditions. With regards to your other requests, galvanic skin response and heart rate changes have been used in the past with polygraphs. These are generally to reconsidered to be poor measures of lie detection as well.Another report here. Although some reports find remarkable findings, such as 80% lie detection, or if you are in china then they a reporting 95% lie detection. Although polygraph was debunked years ago to be at around 60%-65% accurate, see here for an excellent review. As the review points out false positives are a huge problem in polygraph techniques, as they are in other techniques too.