Consider caching findCaseInsensitiveFieldMatch

[jobarr-amzn]

When case insensitivity is enabled the serde may iterate through each struct looking for case-insensitive field matches. A cache in findCaseInsensitiveFieldMatch could speed up this process, reducing struct scans.

https://github.com/amzn/ion-hive-serde/blob/c37e29944b0f25fe08551a6c716a61a3c64181d0/serde/src/main/java/com/amazon/ionhiveserde/caseinsensitivedecorator/IonStructCaseInsensitiveDecorator.java#L125-L132

Would it be worth it? Worth looking at at least.