(0.14.1) https_total_latencies_bucket value discrepancy between Grafana and GCP

[Waizzzzz]

Hello, recently we noticed that values exported by the exporter and the ones in GCP don't match, the metric labels (including bucket lengths) match but the values themselves don't, even when identical queries are used within GCP.
Visually the graphs sometimes match (even with wrong values exported) when there are spikes but the data itself still isn't correct.

Query used in Grafana:
histogram_quantile(0.99,sum by (le)(increase(stackdriver_https_lb_rule_loadbalancing_googleapis_com_https_total_latencies_bucket{target_proxy_name=~"**REDACTED**"}[1m])))

Query used in GCP:
histogram_quantile(0.99,sum by **(le)(increase(loadbalancing_googleapis_com:https_total_latencies_bucket{target_proxy_name=~"REDACTED"}[1m])))