As identified by international telecommunication union (ITU), extremely large-scale multiple-input multiple-output (XL-MIMO) is one of the pivotal enablers for sixth generation (6G) mobile communication networks across mmWave, sub-THz, and emerging mid-band (FR3) frequencies. These systems employ very large antenna arrays to achieve high spatial resolution and throughput. However, the conventional phase-shift-based beamforming which is designed under the narrowband assumption faces limitations due to large bandwidth and array sizes, which subsequently lead to beam squint. This phenomenon exists in both near-field and far-field regions of the antenna. This frequency-dependent misalignment of the beam degrades array gain, increases bit error rate, and limits data rates, especially in wideband scenarios. This paper presents both a comprehensive survey and a unifying framework for beam squint mitigation in XL-MIMO systems. Motivated by the challenges of maintaining consistent beamforming across frequency, we categorize existing solutions into two main classes: subband division-based methods, which are based on filtering the wideband signal to narrowband signals and applying a subband-dependent set of phase shifts to each subband to reduce the beam squint, and subarray division-based techniques, which involve partitioning the entire array into smaller subarrays or utilizing a subarray of true-time-delays (TTDs). We compare these solutions and outline their advantages and disadvantages. Our approach includes theoretical derivation of the beam squint channel for both the far-field and near-field channels, comparative analysis of the beam squint mitigation approaches, and numerical simulations. Finally, we outline open research challenges and suggest potential directions for future work in beam squint mitigation within XL-MIMO systems.