Use `Utf8JsonReader.ValueTextEquals` instead of bytes comparison

[ycrumeyrolle]

Using Utf8JsonReader.ValueTextEquals() cause a drop of performance of about 400% compared to bytes comparison with the signature algorithm scenario. Using Encoding.UTF8.GetBytes(Utf8JsonReader.GetString()) cause a drop of performance of about 80% and an useless allocation of the string object. Using Utf8JsonReader.ValueSpan is fastest, but the value is not unescaped. If the header is represented as {"alg":"HS256"}it is fine, but if it is represented as {"alg":"\u0048\u0053\u0032\u0035\u0036"}, which is not very likely but possible, it will not be correctly interpreted.

An option would be to combine Utf8JsonReader.ValueSpan for most common scenario, and as fallback use Encoding.UTF8.GetBytes(Utf8JsonReader.GetString()). Another option would be to have a API like Utf8JsonReader.GetUtf8String() for extracting the UTF8 unescaped string as its byted representation, similar to the UTF8JsonReader.GetBytesFromBase64() API, but it may still cause some memory allocation. CC @ahsonkhan

Method	data	Mean	Error	StdDev	Median	Ratio	RatioSD	Gen 0	Gen 1	Gen 2	Allocated
Utf8JsonReader.ValueTextEquals	{"alg":"HS256"}	62.035 ns	1.2239 ns	2.2070 ns	62.457 ns	4.86	0.17	-	-	-	-
Utf8JsonReader.ValueTextEquals	{"alg":"FAKE"}	64.928 ns	1.5411 ns	1.5825 ns	64.911 ns	5.28	0.20	-	-	-	-
Encodin_UTF8_GetBytes_Utf8JsonReader_GetString	{"alg":"HS256"}	12.305 ns	0.2001 ns	0.1671 ns	12.298 ns	1.00	0.00	0.0010	-	-	4 B
Encodin_UTF8_GetBytes_Utf8JsonReader_GetString	{"alg":"FAKE"}	12.354 ns	0.2425 ns	0.2886 ns	12.287 ns	1.01	0.04	0.0010	-	-	4 B
Utf8JsonReader_ValueSpan	{"alg":"HS256"}	6.898 ns	0.1436 ns	0.3385 ns	6.731 ns	0.60	0.02	-	-	-	-
Utf8JsonReader_ValueSpan	{"alg":"FAKE"}	6.862 ns	0.1092 ns	0.1073 ns	6.844 ns	0.56	0.01	-	-	-	-