Chunked encoding in tree summary

[daesunp]

Description

This PR changes the sharedTree summary to use the chunked forest codec.

[msfluid-bot]

⯅ @fluid-example/bundle-size-tests: +11.45 KB

Metric Name	Baseline Size	Compare Size	Size Diff
aqueduct.js	447.51 KB	447.51 KB	■ No change
connectionState.js	680 Bytes	680 Bytes	■ No change
containerRuntime.js	240.61 KB	240.61 KB	■ No change
loader.js	178.94 KB	178.94 KB	■ No change
map.js	48.06 KB	48.06 KB	■ No change
matrix.js	141.84 KB	141.84 KB	■ No change
odspDriver.js	90.27 KB	90.27 KB	■ No change
odspPrefetchSnapshot.js	41.9 KB	41.9 KB	■ No change
sharedString.js	162.75 KB	162.75 KB	■ No change
sharedTree2.js	263.51 KB	274.96 KB	⯅ +11.45 KB
Total Size	1.72 MB	1.73 MB	⯅ +11.45 KB

---

Baseline commit: ed4ffda

Generated by 🚫 dangerJS against a88b6f8

[CraigMacomber]

this conversion is unneeded. Since you just need a cursor, decode(content).cursor() can be used as the Delta's content below, and you don't need to generate a jsonable tree then make a cursor from that.

[daesunp]

I have tried to prevent generating a jsonable tree to convert back to a cursor by changing to the following (shown below), but for some reason this does not convert the field cursor into an array of node cursors and the resulting cursors are still in fields mode (which leads to an error).

const cursors = mapCursorField( decode(content).cursor(), (cursor) => cursor as ITreeCursorSynchronous )

However, if I convert to a jsonable tree first (shown below), the cursors seem to be properly converted and it works.

const cursors = mapCursorField( decode(content).cursor(), jsonableTreeFromCursor, ).map(singleTextCursor);

Would there be a better way to convert the field cursor into an array of node cursors?

[daesunp]

updated to const cursors = mapCursorField(decode(content).cursor(), (cursor) => cursor.fork());

[CraigMacomber]

As we store textual diffs for these, if they are going to explicitly an option, I think uncompressed would make more sense

[daesunp]

Updated to TreeCompressionStrategy.Uncompressed

[CraigMacomber]

I think using unreachableCase here would make this code less likley to get out of date if we added more options.

[daesunp]

Updated default case to unreachableCase