[chuuminggg]

prettyPrint() and beneathPath(…) read JSON with an ObjectMapper that uses double for floating-point numbers. When the JSON is written back out, some numbers are documented with a different value or representation from the one in the actual request or response:

{"price":10.50,"rate":1e3,"exact":0.1000000000000000055511151231257827,"big":12345678901234567890.123}

becomes:

{
  "price" : 10.5,
  "rate" : 1000.0,
  "exact" : 0.1,
  "big" : 1.2345678901234567E19
}

The same happens to the subsection that's extracted by beneathPath(…), which is used for the body and fields snippets. Trailing zeros are common in monetary amounts (10.50), so the documented payload can easily differ from what the API actually returns. High-precision and large decimal numbers lose precision.

This PR enables DeserializationFeature.USE_BIG_DECIMAL_FOR_FLOATS on the ObjectMapper that's used for reading in PrettyPrintingContentModifier and FieldPathPayloadSubsectionExtractor, so the numbers are kept as BigDecimal. All of the values above are then written exactly as they were in the input. The only remaining differences are in notation, which doesn't change the value:

• A number in exponent notation is written as BigDecimal.toString() would write it, for example 1E+3 rather than 1e3. Previously, it became 1000.0.
• -0.0 is written as 0.0.

Integers aren't affected, as they were already read as int, long, or BigInteger.