Replace copyContainer with viewContainerVersion

[ChumpChief]

The copyContainer API did not properly handle scenarios where the sequenceNumber from the old container was present in the summary that was used to establish the new container. In this case, the document was corrupted due to a "sequenceNumber from the future" breaking expectations. In 2.0, there is additionally an assert that blocks attempting the copy altogether (added in #17706).

This PR moves away from the copyContainer approach in favor of just loading the historic version for viewing - the customer can then decide what to do with the data. Most likely usage pattern would be:

1. Load the version for viewing
2. Create a new container with the same schema
3. Iterate over the old container's data, writing it into the new container
4. Attach the new container

AB#6411
AB#7712

[msfluid-bot]

⯆ @fluid-example/bundle-size-tests: -104 Bytes

Metric Name	Baseline Size	Compare Size	Size Diff
aqueduct.js	454.11 KB	454.11 KB	■ No change
azureClient.js	550.72 KB	550.62 KB	⯆ -104 Bytes
connectionState.js	680 Bytes	680 Bytes	■ No change
containerRuntime.js	257.28 KB	257.28 KB	■ No change
fluidFramework.js	344.3 KB	344.3 KB	■ No change
loader.js	133.19 KB	133.19 KB	■ No change
map.js	41.44 KB	41.44 KB	■ No change
matrix.js	143.52 KB	143.52 KB	■ No change
odspClient.js	519.1 KB	519.1 KB	■ No change
odspDriver.js	97.52 KB	97.52 KB	■ No change
odspPrefetchSnapshot.js	42.39 KB	42.39 KB	■ No change
sharedString.js	161.25 KB	161.25 KB	■ No change
sharedTree.js	344.29 KB	344.29 KB	■ No change
Total Size	3.17 MB	3.17 MB	⯆ -104 Bytes

---

Baseline commit: eda9e6d

Generated by 🚫 dangerJS against 8b521e4

[scottn12]

LGTM!

[sashasimic]

Is "version" being used at all?

[ChumpChief]

Good call, I accidentally broke it when switching to the new API in the last commit. Fixed in latest.

[sashasimic]

@ChumpChief were you able to validate full recovery flow?

Also please open the ADO item that tracks work needed to updated service docs showing the recovery workflow. Maybe we need to chat with them how their docs are distinguishing 1.x and 2.x scenarios.

[sashasimic]

Would this be a major change?

[ChumpChief]

I believe @tylerbutler 's guidance was to just leave as minor since we don't use changeset to version our packages.

[ChumpChief]

@ChumpChief were you able to validate full recovery flow?

Yes, I did manual verification that I could get versions, use the new API to load from those versions, and retrieve the data (and that the data was in the prior state). EDIT: Also just retested this using the new loadContainerPaused API and confirmed it's all still working as expected.

Also please open the ADO item that tracks work needed to updated service docs showing the recovery workflow. Maybe we need to chat with them how their docs are distinguishing 1.x and 2.x scenarios.

Opened AB#7917