Reduce e2e-test flakiness

[RonFed]

Some of the e2e tests (mainly the kafka-go one) are flaky. For example: https://github.com/open-telemetry/opentelemetry-go-instrumentation/actions/runs/10131243094/job/28013685717.
The cause for this failure is inconsistancy in the batch being exported. Since the default batch timeout is 5s, there is a probability of the spans being sent in 2 different batches/ single batch - which cause the differen JSON resullt in the test.

Proposed solutions:

• The more straightforwrd solution is to define sample-job.yaml per e2e test. Each test will be able to configure the batch span processor using OTEL_BSP_MAX_EXPORT_BATCH_SIZE and OTEL_BSP_SCHEDULE_DELAY accoridng to the expected number of spans in the test.
• A better solution might be to add a number_of_spans field to the test matrix defined in

  opentelemetry-go-instrumentation/.github/workflows/kind.yml

  Line 14 in c80c3ab

  matrix:

  .
  This field will be used to configure OTEL_BSP_MAX_EXPORT_BATCH_SIZE per test. This will require to pass this config to sample-job. It might be possible through a build argument to the Dockerfile, and there are maybe more options.