refactor(tree): Optimize certain leaf access cases

[alexvy86]

Description

Refined PR based on the discussion in #21523 and offline chats, that works towards optimizing access to leaf values in SharedTree.

A couple of comments suggesting that the unboxing behavior of FieldNodes were causing confusion and making this seem more complicated than it ended up being, but it turns out that @CraigMacomber had already removed unboxing from FieldNodes in #18260, so this PR removes those comments and moves ahead with an initial optimization (for some cases) when accessing leaf values.

Reviewer Guidance

The review process is outlined on this wiki page.

The following div shows benchmark results from main and from this PR. Some unhydrated node cases seem to have a bit of a regression, but overall the changes seem good, particularly accessing values that are "simple leaves" (not unions) both in the unhydrated node and flex node cases. For access to leaves (simple and in unions) on object properties, all cases improved.

AB#6908

[alexvy86]

Updated description with latest benchmarks comparing main against the current state of the PR.

[msfluid-bot]

⯅ @fluid-example/bundle-size-tests: +139 Bytes

Metric Name	Baseline Size	Compare Size	Size Diff
aqueduct.js	451.97 KB	452 KB	⯅ +35 Bytes
azureClient.js	550.55 KB	550.59 KB	⯅ +49 Bytes
connectionState.js	680 Bytes	680 Bytes	■ No change
containerRuntime.js	255.81 KB	255.82 KB	⯅ +14 Bytes
fluidFramework.js	384.76 KB	384.72 KB	⯆ -39 Bytes
loader.js	133.46 KB	133.47 KB	⯅ +14 Bytes
map.js	41.1 KB	41.11 KB	⯅ +7 Bytes
matrix.js	143.38 KB	143.38 KB	⯅ +7 Bytes
odspClient.js	518.62 KB	518.66 KB	⯅ +49 Bytes
odspDriver.js	97.21 KB	97.23 KB	⯅ +21 Bytes
odspPrefetchSnapshot.js	42.21 KB	42.22 KB	⯅ +14 Bytes
sharedString.js	159.88 KB	159.89 KB	⯅ +7 Bytes
sharedTree.js	375.22 KB	375.18 KB	⯆ -46 Bytes
Total Size	3.23 MB	3.23 MB	⯅ +139 Bytes

---

Baseline commit: 4fb764f

Generated by 🚫 dangerJS against ead2d3a

[alexvy86]

/azp run Build - client packages

[azure-pipelines]

Azure Pipelines successfully started running 1 pipeline(s).