Standardize docker memory limit for all performance targets

[coderabbitai]

Context

This issue tracks the work to standardize docker memory limits across all performance workflow targets.

Currently, each team's *-performance.yml workflow can specify its own docker_memory value, which leads to inconsistency. The goal is to define a single, sensible default for everyone.

Problem

• Running picofuzz 10 times in a row can cause some targets to go OOM at lower limits (e.g. 512M).
• Some targets have been setting their own values (e.g. 8192m) based on their own fuzzing setup, rather than a shared standard.

Proposal

• Set a fixed docker_memory default of 2048m for all targets in the reusable workflow.
• Remove per-team docker_memory overrides.
• Teams whose targets exceed this limit should optimize their code to work within the 2048m boundary.

References

• PR: Add jampy target to workflows #6
• Comment: Add jampy target to workflows #6 (comment)

/cc @tomusdrw