Change: Improve rate limit algorithm for requests against the NVT API

The original algorithm was very simple by just applying a pause after a
specific number of allowed requests. When doing additional costly tasks
after each requests this algorithm results in unnecessary delays and the
sleep might me completely obsolete. Therefore calculate the time since
the last check for a sleep. If this time delta is within the rate limit
we wait for the required time before doing the next request.

pontos/nvd/api.py

@@ -16,6 +16,7 @@
 # along with this program.  If not, see <http://www.gnu.org/licenses/>.
 
 import asyncio
+import time
 from abc import ABC
 from datetime import datetime, timezone
 from types import TracebackType
@@ -76,10 +77,6 @@ def convert_camel_case(dct: Dict[str, Any]) -> Dict[str, Any]:
     return converted
 
 
-async def sleep() -> None:
-    await asyncio.sleep(SLEEP_TIMEOUT)
-
-
 class NVDApi(ABC):
     """
     Abstract base class for querying the NIST NVD API.
@@ -120,6 +117,7 @@ def __init__(
             self._rate_limit = None
 
         self._request_count = 0
+        self._last_sleep = time.monotonic()
 
     def _request_headers(self) -> Headers:
         """
@@ -141,7 +139,13 @@ async def _consider_rate_limit(self) -> None:
 
         self._request_count += 1
         if self._request_count > self._rate_limit:
-            await sleep()
+            time_since_last_sleep = time.monotonic() - self._last_sleep
+
+            if time_since_last_sleep < SLEEP_TIMEOUT:
+                time_to_sleep = SLEEP_TIMEOUT - time_since_last_sleep
+                await asyncio.sleep(time_to_sleep)
+
+            self._last_sleep = time.monotonic()
             self._request_count = 0
 
     async def _get(