Note extrapolation done in the rate, delta, and increase

As suggested in prometheus/prometheus#2683.

content/docs/querying/functions.md

@@ -89,7 +89,8 @@ month for each of the given times in UTC. Returned values are from 28 to 31.
 `delta(v range-vector)` calculates the difference between the
 first and last value of each time series element in a range vector `v`,
 returning an instant vector with the given deltas and equivalent labels.
-The delta is interpolated to cover the full time range.
+The delta is interpolated to cover the full time range, so that it is 
+possible to get non-integer result even if the samples are all integers.
 
 The following example expression returns the difference in CPU temperature
 between now and 2 hours ago:
@@ -202,7 +203,10 @@ equivalent labels.
 
 `increase(v range-vector)` calculates the increase in the
 time series in the range vector. Breaks in monotonicity (such as counter
-resets due to target restarts) are automatically adjusted for.
+resets due to target restarts) are automatically adjusted for. The
+increment is interpolated to cover the full time range, so that it is
+possible to get non-integer result even if counter increments only by
+integers.
 
 The following example expression returns the number of HTTP requests as measured
 over the last 5 minutes, per time series in the range vector:
@@ -212,8 +216,9 @@ increase(http_requests_total{job="api-server"}[5m])
 ```
 
 `increase` should only be used with counters. It should be used primarily for
-human readability. Use `rate` in recording rules so that increases are tracked
-consistently on a per-second basis.
+human readability and is really just a syntactic sugar around `rate` times the
+number of seconds in the interval. Use `rate` in recording rules so that
+increases are tracked consistently on a per-second basis.
 
 ## `irate()`
 
@@ -299,7 +304,9 @@ regression](http://en.wikipedia.org/wiki/Simple_linear_regression).
 
 `rate(v range-vector)` calculates the per-second average rate of increase of the
 time series in the range vector. Breaks in monotonicity (such as counter
-resets due to target restarts) are automatically adjusted for.
+resets due to target restarts) are automatically adjusted for. Also, the
+calculation extrapolates to the ends of the time range, allowing for missed
+scrapes or imperfect alignment of scrape cycle with the range's time period.
 
 The following example expression returns the per-second rate of HTTP requests as measured
 over the last 5 minutes, per time series in the range vector: