If they missed this, this puts into question all the rest of the data IMO.
"For the years leading up to 1850 we use PAGES2k Consortium reconstruction data. It is based on models where temperatures are reconstructed from proxies. Proxy analysis has higher uncertainty, and we display the smoothed set to highlight the longer-term fluctuations."
PAGES2K has no credibility unfortunately. If the proxies worked they could just use them for the modern era too and avoid the splicing, with thermometer readings providing only greater detail (which isn't important anyway because climate change is about long term trends).
They don't do this. The main reason is because the proxies fail totally in the modern era, with the computed temperatures being very different to observations. The correct interpretation of this is that the chosen proxies don't work for any era, but what they do instead is sweep this fact under the carpet by replacing modern proxy reconstructions with measurements so you can't spot the divergence.
The proxy timeseries also frequently contradict each other in magnitude and direction. For example many proxies show no change over time. If these were truly proxies for global or regional temperature as claimed then different proxies would agree with each other.
If you look at how PAGES2K was constructed it is a festival of pseudo-science. All the usual tricks are there. They delete or truncate data they don't like. What datasets they include varies wildly from release to release without justification. They include tree proxies that they know are distorted by increased CO2=greening in the modern era, and then claim it's a proxy for temperature (this is how Mann got his hockey stick graph in the early 2000s). They even flipped one proxy upside down, the correct interpretation was that temperature had fallen sharply at that location but because they already know what they expect to see, this was mistakenly interpreted backwards and turned into evidence of warming (see the dispute over the varve cores from Hvitavatn in Iceland).
Proxy reconstructions of the past are unfortunately quite a mess.