Stop overriding ObjectMapper naming strategy in Lambda recorder

[tiwari91]

Fixes: #52985

AmazonLambdaMapperRecorder.initObjectMapper() unconditionally called .setPropertyNamingStrategy(PropertyNamingStrategies.LOWER_CAMEL_CASE) on the ObjectMapper used by the Lambda runtime. This silently overrode any application-configured naming strategy (e.g., SNAKE_CASE or KEBAB_CASE), breaking serialization consistency for user DTOs.

Changes

• Removed the .setPropertyNamingStrategy(LOWER_CAMEL_CASE) call from initObjectMapper(), so the application's configured naming strategy is preserved.
• Added @JsonNaming(PropertyNamingStrategies.LowerCamelCaseStrategy.class) to FunctionError, the only internal runtime class serialized by this mapper that lacks explicit @JsonProperty/@JsonGetter annotations. This ensures error reports to the Lambda runtime always use the expected errorType/errorMessage field names regardless of the application's naming strategy.
• Added ObjectMapperNamingStrategyTest verifying that FunctionError always serializes as camelCase even with a SNAKE_CASE mapper, and that user DTOs respect their configured naming strategy.

Why other internal classes don't need changes

• ClientContextImpl and ClientImpl already have explicit @JsonGetter/@JsonSetter annotations with hardcoded field names.
• CognitoIdentity (AWS SDK interface) getter names produce camelCase by default, matching the AWS JSON format.