fix: make LRUCache thread safe

[honzap]

Requirements

• I have added test coverage for new or changed functionality
• I have followed the repository's pull request submission guidelines
• I have validated my changes against all supported platform versions

Describe the solution you've provided

We put LD libraries under the heavy load in production (>500M requests/24h) and the LRU cache starts panics, because it's not thread safe.

We're using the older version of library but the same code is there since a while.

Stack trace (at the bottom, there is a line 32 of lru_cache.go which produces a problem)

/go/pkg/mod/gopkg.in/launchdarkly/go-sdk-events.v1@v1.1.1/event_processor.go:145 +0x616
created by gopkg.in/launchdarkly/go-sdk-events%2ev1.startEventDispatcher in goroutine 525
/go/pkg/mod/gopkg.in/launchdarkly/go-sdk-events.v1@v1.1.1/event_processor.go:193 +0x4c9
gopkg.in/launchdarkly/go-sdk-events%2ev1.(*eventDispatcher).runMainLoop(0xc000b585a0, 0xc000ee8a80)
/go/pkg/mod/gopkg.in/launchdarkly/go-sdk-events.v1@v1.1.1/event_processor.go:263 +0x4ff
gopkg.in/launchdarkly/go-sdk-events%2ev1.(*eventDispatcher).processEvent(0xc000b585a0, {0x49834c0, 0xc0013bf408})
/go/pkg/mod/gopkg.in/launchdarkly/go-sdk-events.v1@v1.1.1/lru_cache.go:32 +0x45
gopkg.in/launchdarkly/go-sdk-events%2ev1.(*lruCache).add(0xc000b58728, {0x37e48a0, 0xc008de97f0})
goroutine 543 [running]: [signal SIGSEGV: segmentation violation code=0x1 addr=0x0 pc=0x476efd]
panic: runtime error: invalid memory address or nil pointer dereference

We think, there should be a RWMutex around that to avoid accessing invalid memory.

Describe alternatives you've considered

No other options.

Additional context

[louis-launchdarkly]

Hello @honzap, I saw @keelerm84 already replied to you in the GitHub issue, and he can help in looking at this contribution.

Filed Internally as SDK-1099