reduce golangci-lint overhead for jenkins worker

The crossplane jenkins worker node only has 10gb of RAM.  With 1.17.1 of
golangci-lint we regularly see 10gb consumed while linting in local
environments.

The changes suggested here appear to improve the overhead.
golangci/golangci-lint#337 (comment)

Prior to the bump to 1.17.1, golangci-lint was taking 6-8GB of RAM and
completing in 10s.

With the reduction in overhead to golangci-lint that this PR introduces,
linting consumes between 8-9GB of ram and completes in 30s using the same
environment.

The intent here is to continue using latest golangci-lint with the
current Jenkins configuration, sacrificing some time.  Additional
measure can be made, such as using `--fast` tests only or further increasing the
LINT_GOGC value.

https://github.com/golangci/golangci-lint#configuration
https://github.com/golangci/golangci-lint#memory-usage-of-golangci-lint

Signed-off-by: Marques Johansson <[email protected]>

Author: displague

makelib/golang.mk

@@ -117,8 +117,10 @@ GO_OUT_EXT := .exe
 endif
 
 ifeq ($(RUNNING_IN_CI),true)
+# Reduce concurrency to reduce RAM requirements on jenkins worker nodes
+# increase deadline to 3m (from default 1m, crossplane default 2m) to potentially compensate for less concurrency
 # Output checkstyle XML rather than human readable output.
-GO_LINT_ARGS := --out-format=checkstyle > $(GO_LINT_OUTPUT)/checkstyle.xml
+GO_LINT_ARGS := --concurrency=1 --deadline=3m0s --out-format=checkstyle > $(GO_LINT_OUTPUT)/checkstyle.xml
 
 # Output verbose tests that can be parsed into JUnit XML.
 GO_TEST_FLAGS += -v
@@ -192,7 +194,7 @@ go.test.integration: $(GOJUNIT)
 go.lint: $(GOLANGCILINT)
 	@$(INFO) golangci-lint
 	@mkdir -p $(GO_LINT_OUTPUT)
-	@$(GOLANGCILINT) run $(GO_LINT_ARGS) || $(FAIL)
+	@LINT_GOGC=20 $(GOLANGCILINT) run $(GO_LINT_ARGS) || $(FAIL)
 	@$(OK) golangci-lint
 
 go.vet: