stdlib: Sprinkle an __always inline onto Range.index(after:)

[aschwaighofer]

After moving to the new llvm pass pipeline this function no longer gets
inlined. Leading to 150% regressions in ObjectiveCBridge benchmarks.

rdar://98533158

[aschwaighofer]

@swift-ci benchmark

[aschwaighofer]

From the benchmark run:

IMPROVEMENT                                        OLD    NEW    DELTA    RATIO    
Breadcrumbs.IdxToUTF16Range.longMixed              2794   401    -85.6%   **6.97x**
Breadcrumbs.IdxToUTF16.longMixed                   4761   733    -84.6%   **6.50x**
ObjectiveCBridgeStringIsEqualAllSwift              130    52     -60.0%   **2.50x**
ObjectiveCBridgeStringIsEqual                      368    148    -59.8%   **2.49x**
ObjectiveCBridgeStringIsEqual2                     405    169    -58.3%   **2.40x**
ObjectiveCBridgeStringGetASCIIContents             591    261    -55.8%   **2.26x**
ObjectiveCBridgeStringHash                         155    69     -55.5%   **2.25x**
ObjectiveCBridgeStringCStringUsingEncoding         899    477    -46.9%   **1.88x**
ObjectiveCBridgeStringRangeOfString                1124   630    -44.0%   **1.78x**
NSStringConversion.MutableCopy.Rebridge.LongUTF8   601    355    -40.9%   **1.69x**
BridgeString.find.native.nonASCII                  463    303    -34.6%   **1.53x**
ObjectiveCBridgeStringCompare2                     925    610    -34.1%   **1.52x**
ObjectiveCBridgeStringCompare                      951    633    -33.4%   **1.50x**
NSStringConversion.MutableCopy.LongUTF8            489    372    -23.9%   **1.31x**

[aschwaighofer]

@swift-ci test

[Catfish-Man]

Looks good to me, thank you!